; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+zfbfmin,+zvfbfmin,+v \
; RUN: -target-abi=ilp32d -verify-machineinstrs < %s | FileCheck %s \
; RUN: --check-prefixes=CHECK,ZVFH
; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+zfbfmin,+zvfbfmin,+v \
; RUN: -target-abi=lp64d -verify-machineinstrs < %s | FileCheck %s \
; RUN: --check-prefixes=CHECK,ZVFH
; RUN: llc -mtriple=riscv32 -mattr=+d,+zfhmin,+zvfhmin,+zfbfmin,+zvfbfmin,+v \
; RUN: -target-abi=ilp32d -verify-machineinstrs < %s | FileCheck %s \
; RUN: --check-prefixes=CHECK,ZVFHMIN
; RUN: llc -mtriple=riscv64 -mattr=+d,+zfhmin,+zvfhmin,+zfbfmin,+zvfbfmin,+v \
; RUN: -target-abi=lp64d -verify-machineinstrs < %s | FileCheck %s \
; RUN: --check-prefixes=CHECK,ZVFHMIN
declare <vscale x 1 x bfloat> @llvm.vp.fma.nxv1bf16(<vscale x 1 x bfloat>, <vscale x 1 x bfloat>, <vscale x 1 x bfloat>, <vscale x 1 x i1>, i32)
define <vscale x 1 x bfloat> @vfma_vv_nxv1bf16(<vscale x 1 x bfloat> %va, <vscale x 1 x bfloat> %b, <vscale x 1 x bfloat> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv1bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v10
; CHECK-NEXT: vfwcvtbf16.f.f.v v10, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v9
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v10, v11, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v12
; CHECK-NEXT: ret
%v = call <vscale x 1 x bfloat> @llvm.vp.fma.nxv1bf16(<vscale x 1 x bfloat> %va, <vscale x 1 x bfloat> %b, <vscale x 1 x bfloat> %c, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x bfloat> %v
}
define <vscale x 1 x bfloat> @vfma_vv_nxv1bf16_unmasked(<vscale x 1 x bfloat> %va, <vscale x 1 x bfloat> %b, <vscale x 1 x bfloat> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv1bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v10
; CHECK-NEXT: vfwcvtbf16.f.f.v v10, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v9
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v10, v11
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v12
; CHECK-NEXT: ret
%v = call <vscale x 1 x bfloat> @llvm.vp.fma.nxv1bf16(<vscale x 1 x bfloat> %va, <vscale x 1 x bfloat> %b, <vscale x 1 x bfloat> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x bfloat> %v
}
define <vscale x 1 x bfloat> @vfma_vf_nxv1bf16(<vscale x 1 x bfloat> %va, bfloat %b, <vscale x 1 x bfloat> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v9, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v10
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v9, v11, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 1 x bfloat> %elt.head, <vscale x 1 x bfloat> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x bfloat> @llvm.vp.fma.nxv1bf16(<vscale x 1 x bfloat> %va, <vscale x 1 x bfloat> %vb, <vscale x 1 x bfloat> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x bfloat> %v
}
define <vscale x 1 x bfloat> @vfma_vf_nxv1bf16_commute(<vscale x 1 x bfloat> %va, bfloat %b, <vscale x 1 x bfloat> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1bf16_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v9, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v10
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vv v9, v8, v11, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 1 x bfloat> %elt.head, <vscale x 1 x bfloat> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x bfloat> @llvm.vp.fma.nxv1bf16(<vscale x 1 x bfloat> %vb, <vscale x 1 x bfloat> %va, <vscale x 1 x bfloat> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x bfloat> %v
}
define <vscale x 1 x bfloat> @vfma_vf_nxv1bf16_unmasked(<vscale x 1 x bfloat> %va, bfloat %b, <vscale x 1 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v9, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v10
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v9, v11
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 1 x bfloat> %elt.head, <vscale x 1 x bfloat> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x bfloat> @llvm.vp.fma.nxv1bf16(<vscale x 1 x bfloat> %va, <vscale x 1 x bfloat> %vb, <vscale x 1 x bfloat> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x bfloat> %v
}
define <vscale x 1 x bfloat> @vfma_vf_nxv1bf16_unmasked_commute(<vscale x 1 x bfloat> %va, bfloat %b, <vscale x 1 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1bf16_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v9, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v10
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v9, v11
; CHECK-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 1 x bfloat> %elt.head, <vscale x 1 x bfloat> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x bfloat> @llvm.vp.fma.nxv1bf16(<vscale x 1 x bfloat> %vb, <vscale x 1 x bfloat> %va, <vscale x 1 x bfloat> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x bfloat> %v
}
declare <vscale x 2 x bfloat> @llvm.vp.fma.nxv2bf16(<vscale x 2 x bfloat>, <vscale x 2 x bfloat>, <vscale x 2 x bfloat>, <vscale x 2 x i1>, i32)
define <vscale x 2 x bfloat> @vfma_vv_nxv2bf16(<vscale x 2 x bfloat> %va, <vscale x 2 x bfloat> %b, <vscale x 2 x bfloat> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv2bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v10
; CHECK-NEXT: vfwcvtbf16.f.f.v v10, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v9
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v10, v11, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v12
; CHECK-NEXT: ret
%v = call <vscale x 2 x bfloat> @llvm.vp.fma.nxv2bf16(<vscale x 2 x bfloat> %va, <vscale x 2 x bfloat> %b, <vscale x 2 x bfloat> %c, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x bfloat> %v
}
define <vscale x 2 x bfloat> @vfma_vv_nxv2bf16_unmasked(<vscale x 2 x bfloat> %va, <vscale x 2 x bfloat> %b, <vscale x 2 x bfloat> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv2bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v10
; CHECK-NEXT: vfwcvtbf16.f.f.v v10, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v9
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v10, v11
; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v12
; CHECK-NEXT: ret
%v = call <vscale x 2 x bfloat> @llvm.vp.fma.nxv2bf16(<vscale x 2 x bfloat> %va, <vscale x 2 x bfloat> %b, <vscale x 2 x bfloat> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x bfloat> %v
}
define <vscale x 2 x bfloat> @vfma_vf_nxv2bf16(<vscale x 2 x bfloat> %va, bfloat %b, <vscale x 2 x bfloat> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v9, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v10
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v9, v11, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 2 x bfloat> %elt.head, <vscale x 2 x bfloat> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x bfloat> @llvm.vp.fma.nxv2bf16(<vscale x 2 x bfloat> %va, <vscale x 2 x bfloat> %vb, <vscale x 2 x bfloat> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x bfloat> %v
}
define <vscale x 2 x bfloat> @vfma_vf_nxv2bf16_commute(<vscale x 2 x bfloat> %va, bfloat %b, <vscale x 2 x bfloat> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2bf16_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v9, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v10
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vv v9, v8, v11, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 2 x bfloat> %elt.head, <vscale x 2 x bfloat> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x bfloat> @llvm.vp.fma.nxv2bf16(<vscale x 2 x bfloat> %vb, <vscale x 2 x bfloat> %va, <vscale x 2 x bfloat> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x bfloat> %v
}
define <vscale x 2 x bfloat> @vfma_vf_nxv2bf16_unmasked(<vscale x 2 x bfloat> %va, bfloat %b, <vscale x 2 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v9, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v10
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v9, v11
; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 2 x bfloat> %elt.head, <vscale x 2 x bfloat> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x bfloat> @llvm.vp.fma.nxv2bf16(<vscale x 2 x bfloat> %va, <vscale x 2 x bfloat> %vb, <vscale x 2 x bfloat> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x bfloat> %v
}
define <vscale x 2 x bfloat> @vfma_vf_nxv2bf16_unmasked_commute(<vscale x 2 x bfloat> %va, bfloat %b, <vscale x 2 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2bf16_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v11, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v9, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v10
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v9, v11
; CHECK-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 2 x bfloat> %elt.head, <vscale x 2 x bfloat> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x bfloat> @llvm.vp.fma.nxv2bf16(<vscale x 2 x bfloat> %vb, <vscale x 2 x bfloat> %va, <vscale x 2 x bfloat> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x bfloat> %v
}
declare <vscale x 4 x bfloat> @llvm.vp.fma.nxv4bf16(<vscale x 4 x bfloat>, <vscale x 4 x bfloat>, <vscale x 4 x bfloat>, <vscale x 4 x i1>, i32)
define <vscale x 4 x bfloat> @vfma_vv_nxv4bf16(<vscale x 4 x bfloat> %va, <vscale x 4 x bfloat> %b, <vscale x 4 x bfloat> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv4bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v10
; CHECK-NEXT: vfwcvtbf16.f.f.v v10, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v14, v9
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vv v14, v10, v12, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v14
; CHECK-NEXT: ret
%v = call <vscale x 4 x bfloat> @llvm.vp.fma.nxv4bf16(<vscale x 4 x bfloat> %va, <vscale x 4 x bfloat> %b, <vscale x 4 x bfloat> %c, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x bfloat> %v
}
define <vscale x 4 x bfloat> @vfma_vv_nxv4bf16_unmasked(<vscale x 4 x bfloat> %va, <vscale x 4 x bfloat> %b, <vscale x 4 x bfloat> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv4bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v10
; CHECK-NEXT: vfwcvtbf16.f.f.v v10, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v14, v9
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vv v14, v10, v12
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v14
; CHECK-NEXT: ret
%v = call <vscale x 4 x bfloat> @llvm.vp.fma.nxv4bf16(<vscale x 4 x bfloat> %va, <vscale x 4 x bfloat> %b, <vscale x 4 x bfloat> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x bfloat> %v
}
define <vscale x 4 x bfloat> @vfma_vf_nxv4bf16(<vscale x 4 x bfloat> %va, bfloat %b, <vscale x 4 x bfloat> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v14, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v10
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v14, v12, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 4 x bfloat> %elt.head, <vscale x 4 x bfloat> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x bfloat> @llvm.vp.fma.nxv4bf16(<vscale x 4 x bfloat> %va, <vscale x 4 x bfloat> %vb, <vscale x 4 x bfloat> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x bfloat> %v
}
define <vscale x 4 x bfloat> @vfma_vf_nxv4bf16_commute(<vscale x 4 x bfloat> %va, bfloat %b, <vscale x 4 x bfloat> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4bf16_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v14, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v10
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vv v14, v8, v12, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v14
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 4 x bfloat> %elt.head, <vscale x 4 x bfloat> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x bfloat> @llvm.vp.fma.nxv4bf16(<vscale x 4 x bfloat> %vb, <vscale x 4 x bfloat> %va, <vscale x 4 x bfloat> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x bfloat> %v
}
define <vscale x 4 x bfloat> @vfma_vf_nxv4bf16_unmasked(<vscale x 4 x bfloat> %va, bfloat %b, <vscale x 4 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v14, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v10
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v14, v12
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 4 x bfloat> %elt.head, <vscale x 4 x bfloat> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x bfloat> @llvm.vp.fma.nxv4bf16(<vscale x 4 x bfloat> %va, <vscale x 4 x bfloat> %vb, <vscale x 4 x bfloat> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x bfloat> %v
}
define <vscale x 4 x bfloat> @vfma_vf_nxv4bf16_unmasked_commute(<vscale x 4 x bfloat> %va, bfloat %b, <vscale x 4 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4bf16_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; CHECK-NEXT: vmv.v.x v10, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v9
; CHECK-NEXT: vfwcvtbf16.f.f.v v14, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v10
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v14, v12
; CHECK-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 4 x bfloat> %elt.head, <vscale x 4 x bfloat> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x bfloat> @llvm.vp.fma.nxv4bf16(<vscale x 4 x bfloat> %vb, <vscale x 4 x bfloat> %va, <vscale x 4 x bfloat> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x bfloat> %v
}
declare <vscale x 8 x bfloat> @llvm.vp.fma.nxv8bf16(<vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x i1>, i32)
define <vscale x 8 x bfloat> @vfma_vv_nxv8bf16(<vscale x 8 x bfloat> %va, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv8bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v12
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v20, v10
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vv v20, v12, v16, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v20
; CHECK-NEXT: ret
%v = call <vscale x 8 x bfloat> @llvm.vp.fma.nxv8bf16(<vscale x 8 x bfloat> %va, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x bfloat> %v
}
define <vscale x 8 x bfloat> @vfma_vv_nxv8bf16_unmasked(<vscale x 8 x bfloat> %va, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv8bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v12
; CHECK-NEXT: vfwcvtbf16.f.f.v v12, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v20, v10
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vv v20, v12, v16
; CHECK-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v20
; CHECK-NEXT: ret
%v = call <vscale x 8 x bfloat> @llvm.vp.fma.nxv8bf16(<vscale x 8 x bfloat> %va, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x bfloat> %v
}
define <vscale x 8 x bfloat> @vfma_vf_nxv8bf16(<vscale x 8 x bfloat> %va, bfloat %b, <vscale x 8 x bfloat> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; CHECK-NEXT: vmv.v.x v12, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v10
; CHECK-NEXT: vfwcvtbf16.f.f.v v20, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v12
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vv v24, v20, v16, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v24
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 8 x bfloat> %elt.head, <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x bfloat> @llvm.vp.fma.nxv8bf16(<vscale x 8 x bfloat> %va, <vscale x 8 x bfloat> %vb, <vscale x 8 x bfloat> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x bfloat> %v
}
define <vscale x 8 x bfloat> @vfma_vf_nxv8bf16_commute(<vscale x 8 x bfloat> %va, bfloat %b, <vscale x 8 x bfloat> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8bf16_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; CHECK-NEXT: vmv.v.x v12, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v10
; CHECK-NEXT: vfwcvtbf16.f.f.v v20, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v12
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vv v20, v8, v16, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v20
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 8 x bfloat> %elt.head, <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x bfloat> @llvm.vp.fma.nxv8bf16(<vscale x 8 x bfloat> %vb, <vscale x 8 x bfloat> %va, <vscale x 8 x bfloat> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x bfloat> %v
}
define <vscale x 8 x bfloat> @vfma_vf_nxv8bf16_unmasked(<vscale x 8 x bfloat> %va, bfloat %b, <vscale x 8 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; CHECK-NEXT: vmv.v.x v12, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v10
; CHECK-NEXT: vfwcvtbf16.f.f.v v20, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v12
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vv v24, v20, v16
; CHECK-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v24
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 8 x bfloat> %elt.head, <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x bfloat> @llvm.vp.fma.nxv8bf16(<vscale x 8 x bfloat> %va, <vscale x 8 x bfloat> %vb, <vscale x 8 x bfloat> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x bfloat> %v
}
define <vscale x 8 x bfloat> @vfma_vf_nxv8bf16_unmasked_commute(<vscale x 8 x bfloat> %va, bfloat %b, <vscale x 8 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8bf16_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; CHECK-NEXT: vmv.v.x v12, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v10
; CHECK-NEXT: vfwcvtbf16.f.f.v v20, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v12
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vv v24, v20, v16
; CHECK-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v24
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 8 x bfloat> %elt.head, <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x bfloat> @llvm.vp.fma.nxv8bf16(<vscale x 8 x bfloat> %vb, <vscale x 8 x bfloat> %va, <vscale x 8 x bfloat> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x bfloat> %v
}
declare <vscale x 16 x bfloat> @llvm.vp.fma.nxv16bf16(<vscale x 16 x bfloat>, <vscale x 16 x bfloat>, <vscale x 16 x bfloat>, <vscale x 16 x i1>, i32)
define <vscale x 16 x bfloat> @vfma_vv_nxv16bf16(<vscale x 16 x bfloat> %va, <vscale x 16 x bfloat> %b, <vscale x 16 x bfloat> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv16bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: sub sp, sp, a1
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 8 * vlenb
; CHECK-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v16
; CHECK-NEXT: addi a1, sp, 16
; CHECK-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v12
; CHECK-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v24, v8, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v16
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%v = call <vscale x 16 x bfloat> @llvm.vp.fma.nxv16bf16(<vscale x 16 x bfloat> %va, <vscale x 16 x bfloat> %b, <vscale x 16 x bfloat> %c, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x bfloat> %v
}
define <vscale x 16 x bfloat> @vfma_vv_nxv16bf16_unmasked(<vscale x 16 x bfloat> %va, <vscale x 16 x bfloat> %b, <vscale x 16 x bfloat> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv16bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v16
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v0, v12
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v0, v16, v24
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v0
; CHECK-NEXT: ret
%v = call <vscale x 16 x bfloat> @llvm.vp.fma.nxv16bf16(<vscale x 16 x bfloat> %va, <vscale x 16 x bfloat> %b, <vscale x 16 x bfloat> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x bfloat> %v
}
define <vscale x 16 x bfloat> @vfma_vf_nxv16bf16(<vscale x 16 x bfloat> %va, bfloat %b, <vscale x 16 x bfloat> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv16bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: sub sp, sp, a1
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 8 * vlenb
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vmv.v.x v4, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v12
; CHECK-NEXT: addi a1, sp, 16
; CHECK-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v4
; CHECK-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v24, v8, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v16
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 16 x bfloat> %elt.head, <vscale x 16 x bfloat> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x bfloat> @llvm.vp.fma.nxv16bf16(<vscale x 16 x bfloat> %va, <vscale x 16 x bfloat> %vb, <vscale x 16 x bfloat> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x bfloat> %v
}
define <vscale x 16 x bfloat> @vfma_vf_nxv16bf16_commute(<vscale x 16 x bfloat> %va, bfloat %b, <vscale x 16 x bfloat> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv16bf16_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vmv.v.x v4, a1
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v12
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v8
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v4
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v8, v24, v0.t
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 16 x bfloat> %elt.head, <vscale x 16 x bfloat> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x bfloat> @llvm.vp.fma.nxv16bf16(<vscale x 16 x bfloat> %vb, <vscale x 16 x bfloat> %va, <vscale x 16 x bfloat> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x bfloat> %v
}
define <vscale x 16 x bfloat> @vfma_vf_nxv16bf16_unmasked(<vscale x 16 x bfloat> %va, bfloat %b, <vscale x 16 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv16bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 2
; CHECK-NEXT: sub sp, sp, a1
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 4 * vlenb
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vmv.v.x v16, a1
; CHECK-NEXT: addi a1, sp, 16
; CHECK-NEXT: vs4r.v v16, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v12
; CHECK-NEXT: vfwcvtbf16.f.f.v v0, v8
; CHECK-NEXT: vl4r.v v8, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v8
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v0, v24
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v16
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 2
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 16 x bfloat> %elt.head, <vscale x 16 x bfloat> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x bfloat> @llvm.vp.fma.nxv16bf16(<vscale x 16 x bfloat> %va, <vscale x 16 x bfloat> %vb, <vscale x 16 x bfloat> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x bfloat> %v
}
define <vscale x 16 x bfloat> @vfma_vf_nxv16bf16_unmasked_commute(<vscale x 16 x bfloat> %va, bfloat %b, <vscale x 16 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv16bf16_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 2
; CHECK-NEXT: sub sp, sp, a1
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 4 * vlenb
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vmv.v.x v16, a1
; CHECK-NEXT: addi a1, sp, 16
; CHECK-NEXT: vs4r.v v16, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v12
; CHECK-NEXT: vfwcvtbf16.f.f.v v0, v8
; CHECK-NEXT: vl4r.v v8, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v8
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v0, v24
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v16
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 2
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 16 x bfloat> %elt.head, <vscale x 16 x bfloat> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x bfloat> @llvm.vp.fma.nxv16bf16(<vscale x 16 x bfloat> %vb, <vscale x 16 x bfloat> %va, <vscale x 16 x bfloat> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x bfloat> %v
}
declare <vscale x 32 x bfloat> @llvm.vp.fma.nxv32bf16(<vscale x 32 x bfloat>, <vscale x 32 x bfloat>, <vscale x 32 x bfloat>, <vscale x 32 x i1>, i32)
define <vscale x 32 x bfloat> @vfma_vv_nxv32bf16(<vscale x 32 x bfloat> %va, <vscale x 32 x bfloat> %b, <vscale x 32 x bfloat> %c, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv32bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: mv a3, a2
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a3, a3, a2
; CHECK-NEXT: slli a2, a2, 2
; CHECK-NEXT: add a2, a2, a3
; CHECK-NEXT: sub sp, sp, a2
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x29, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 41 * vlenb
; CHECK-NEXT: vmv1r.v v24, v0
; CHECK-NEXT: vl8re16.v v0, (a0)
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a0, a2, 1
; CHECK-NEXT: sub a3, a1, a0
; CHECK-NEXT: sltu a4, a1, a3
; CHECK-NEXT: addi a4, a4, -1
; CHECK-NEXT: and a3, a4, a3
; CHECK-NEXT: srli a2, a2, 2
; CHECK-NEXT: csrr a4, vlenb
; CHECK-NEXT: slli a4, a4, 4
; CHECK-NEXT: add a4, sp, a4
; CHECK-NEXT: addi a4, a4, 16
; CHECK-NEXT: vs1r.v v24, (a4) # Unknown-size Folded Spill
; CHECK-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; CHECK-NEXT: vslidedown.vx v24, v24, a2
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vs1r.v v24, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: mv a4, a2
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a4, a4, a2
; CHECK-NEXT: slli a2, a2, 1
; CHECK-NEXT: add a2, a2, a4
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v12
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vmv8r.v v8, v16
; CHECK-NEXT: vmv4r.v v24, v8
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a4, a2, 5
; CHECK-NEXT: add a2, a4, a2
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v12
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a4, a2, 4
; CHECK-NEXT: add a2, a4, a2
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v0, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v4
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vl1r.v v0, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v24, v8, v0.t
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v12, v16
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: bltu a1, a0, .LBB30_2
; CHECK-NEXT: # %bb.1:
; CHECK-NEXT: mv a1, a0
; CHECK-NEXT: .LBB30_2:
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: mv a2, a0
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: add a2, a2, a0
; CHECK-NEXT: slli a0, a0, 1
; CHECK-NEXT: add a0, a0, a2
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v16
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vs8r.v v24, (a0) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a2, a0, 5
; CHECK-NEXT: add a0, a2, a0
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v16
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a2, a0, 4
; CHECK-NEXT: add a0, a2, a0
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v0, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v0
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a2, a0, 5
; CHECK-NEXT: add a0, a2, a0
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vs8r.v v16, (a0) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 4
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl1r.v v0, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a2, a0, 5
; CHECK-NEXT: add a0, a2, a0
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v16, v24, v0.t
; CHECK-NEXT: vmv.v.v v16, v8
; CHECK-NEXT: addi a0, sp, 16
; CHECK-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v16
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: mv a1, a0
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: add a1, a1, a0
; CHECK-NEXT: slli a0, a0, 2
; CHECK-NEXT: add a0, a0, a1
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%v = call <vscale x 32 x bfloat> @llvm.vp.fma.nxv32bf16(<vscale x 32 x bfloat> %va, <vscale x 32 x bfloat> %b, <vscale x 32 x bfloat> %c, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x bfloat> %v
}
define <vscale x 32 x bfloat> @vfma_vv_nxv32bf16_unmasked(<vscale x 32 x bfloat> %va, <vscale x 32 x bfloat> %b, <vscale x 32 x bfloat> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv32bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 5
; CHECK-NEXT: sub sp, sp, a2
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; CHECK-NEXT: vmv8r.v v24, v16
; CHECK-NEXT: vl8re16.v v16, (a0)
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: mv a2, a0
; CHECK-NEXT: slli a0, a0, 1
; CHECK-NEXT: add a0, a0, a2
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vs8r.v v16, (a0) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a0, a2, 1
; CHECK-NEXT: sub a3, a1, a0
; CHECK-NEXT: sltu a4, a1, a3
; CHECK-NEXT: addi a4, a4, -1
; CHECK-NEXT: and a3, a4, a3
; CHECK-NEXT: srli a2, a2, 2
; CHECK-NEXT: vsetvli a4, zero, e8, m4, ta, ma
; CHECK-NEXT: vmset.m v7
; CHECK-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; CHECK-NEXT: vslidedown.vx v0, v7, a2
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v12
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 4
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v28
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: mv a4, a2
; CHECK-NEXT: slli a2, a2, 1
; CHECK-NEXT: add a2, a2, a4
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v28
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v24, v8, v0.t
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v12, v16
; CHECK-NEXT: bltu a1, a0, .LBB31_2
; CHECK-NEXT: # %bb.1:
; CHECK-NEXT: mv a1, a0
; CHECK-NEXT: .LBB31_2:
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v16
; CHECK-NEXT: addi a0, sp, 16
; CHECK-NEXT: vs8r.v v24, (a0) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 4
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v0, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v0
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: mv a2, a0
; CHECK-NEXT: slli a0, a0, 1
; CHECK-NEXT: add a0, a0, a2
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v0, v16
; CHECK-NEXT: addi a0, sp, 16
; CHECK-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfmacc.vv v0, v16, v24
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v0
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 5
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%v = call <vscale x 32 x bfloat> @llvm.vp.fma.nxv32bf16(<vscale x 32 x bfloat> %va, <vscale x 32 x bfloat> %b, <vscale x 32 x bfloat> %c, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x bfloat> %v
}
define <vscale x 32 x bfloat> @vfma_vf_nxv32bf16(<vscale x 32 x bfloat> %va, bfloat %b, <vscale x 32 x bfloat> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv32bf16:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a2, a2, a1
; CHECK-NEXT: slli a1, a1, 2
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: sub sp, sp, a1
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x29, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 41 * vlenb
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; CHECK-NEXT: vmv.v.x v24, a1
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a2, a1, 5
; CHECK-NEXT: add a1, a2, a1
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a1, a2, 1
; CHECK-NEXT: sub a3, a0, a1
; CHECK-NEXT: sltu a4, a0, a3
; CHECK-NEXT: addi a4, a4, -1
; CHECK-NEXT: and a3, a4, a3
; CHECK-NEXT: srli a2, a2, 2
; CHECK-NEXT: csrr a4, vlenb
; CHECK-NEXT: slli a4, a4, 4
; CHECK-NEXT: add a4, sp, a4
; CHECK-NEXT: addi a4, a4, 16
; CHECK-NEXT: vs1r.v v0, (a4) # Unknown-size Folded Spill
; CHECK-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; CHECK-NEXT: vslidedown.vx v0, v0, a2
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: mv a4, a2
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a4, a4, a2
; CHECK-NEXT: slli a2, a2, 1
; CHECK-NEXT: add a2, a2, a4
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v20
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a4, a2, 4
; CHECK-NEXT: add a2, a4, a2
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v12
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a4, a2, 5
; CHECK-NEXT: add a2, a4, a2
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v28
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v16, v24, v0.t
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v20, v8
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: bltu a0, a1, .LBB32_2
; CHECK-NEXT: # %bb.1:
; CHECK-NEXT: mv a0, a1
; CHECK-NEXT: .LBB32_2:
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a2, a2, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v8
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a2, a1, 4
; CHECK-NEXT: add a1, a2, a1
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v8
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a2, a2, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a2, a1, 5
; CHECK-NEXT: add a1, a2, a1
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v0
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 4
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl1r.v v0, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a2, a2, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v16, v24, v0.t
; CHECK-NEXT: vmv.v.v v16, v8
; CHECK-NEXT: addi a0, sp, 16
; CHECK-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v16
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: mv a1, a0
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: add a1, a1, a0
; CHECK-NEXT: slli a0, a0, 2
; CHECK-NEXT: add a0, a0, a1
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 32 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 32 x bfloat> %elt.head, <vscale x 32 x bfloat> poison, <vscale x 32 x i32> zeroinitializer
%v = call <vscale x 32 x bfloat> @llvm.vp.fma.nxv32bf16(<vscale x 32 x bfloat> %va, <vscale x 32 x bfloat> %vb, <vscale x 32 x bfloat> %vc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x bfloat> %v
}
define <vscale x 32 x bfloat> @vfma_vf_nxv32bf16_commute(<vscale x 32 x bfloat> %va, bfloat %b, <vscale x 32 x bfloat> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv32bf16_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a2, a2, a1
; CHECK-NEXT: slli a1, a1, 2
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: sub sp, sp, a1
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x29, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 41 * vlenb
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; CHECK-NEXT: vmv.v.x v24, a1
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a2, a1, 5
; CHECK-NEXT: add a1, a2, a1
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a1, a2, 1
; CHECK-NEXT: sub a3, a0, a1
; CHECK-NEXT: sltu a4, a0, a3
; CHECK-NEXT: addi a4, a4, -1
; CHECK-NEXT: and a3, a4, a3
; CHECK-NEXT: srli a2, a2, 2
; CHECK-NEXT: csrr a4, vlenb
; CHECK-NEXT: slli a4, a4, 4
; CHECK-NEXT: add a4, sp, a4
; CHECK-NEXT: addi a4, a4, 16
; CHECK-NEXT: vs1r.v v0, (a4) # Unknown-size Folded Spill
; CHECK-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; CHECK-NEXT: vslidedown.vx v0, v0, a2
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: mv a4, a2
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a4, a4, a2
; CHECK-NEXT: slli a2, a2, 1
; CHECK-NEXT: add a2, a2, a4
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v20
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a4, a2, 4
; CHECK-NEXT: add a2, a4, a2
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v12
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a4, a2, 5
; CHECK-NEXT: add a2, a4, a2
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v28
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v8, v24, v0.t
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v12, v16
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: bltu a0, a1, .LBB33_2
; CHECK-NEXT: # %bb.1:
; CHECK-NEXT: mv a0, a1
; CHECK-NEXT: .LBB33_2:
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a2, a2, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a2, a1, 4
; CHECK-NEXT: add a1, a2, a1
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a2, a1, 5
; CHECK-NEXT: add a1, a2, a1
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v0
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a2, a2, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 4
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl1r.v v0, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a2, a2, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v24, v16, v0.t
; CHECK-NEXT: vmv.v.v v16, v8
; CHECK-NEXT: addi a0, sp, 16
; CHECK-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v16
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: mv a1, a0
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: add a1, a1, a0
; CHECK-NEXT: slli a0, a0, 2
; CHECK-NEXT: add a0, a0, a1
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 32 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 32 x bfloat> %elt.head, <vscale x 32 x bfloat> poison, <vscale x 32 x i32> zeroinitializer
%v = call <vscale x 32 x bfloat> @llvm.vp.fma.nxv32bf16(<vscale x 32 x bfloat> %vb, <vscale x 32 x bfloat> %va, <vscale x 32 x bfloat> %vc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x bfloat> %v
}
define <vscale x 32 x bfloat> @vfma_vf_nxv32bf16_unmasked(<vscale x 32 x bfloat> %va, bfloat %b, <vscale x 32 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv32bf16_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 5
; CHECK-NEXT: sub sp, sp, a1
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; CHECK-NEXT: vmv.v.x v24, a1
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a1, a2, 1
; CHECK-NEXT: sub a3, a0, a1
; CHECK-NEXT: sltu a4, a0, a3
; CHECK-NEXT: addi a4, a4, -1
; CHECK-NEXT: and a3, a4, a3
; CHECK-NEXT: srli a2, a2, 2
; CHECK-NEXT: vmset.m v7
; CHECK-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; CHECK-NEXT: vslidedown.vx v0, v7, a2
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 4
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v20
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v12
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: mv a4, a2
; CHECK-NEXT: slli a2, a2, 1
; CHECK-NEXT: add a2, a2, a4
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v28
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v16, v24, v0.t
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v20, v8
; CHECK-NEXT: bltu a0, a1, .LBB34_2
; CHECK-NEXT: # %bb.1:
; CHECK-NEXT: mv a0, a1
; CHECK-NEXT: .LBB34_2:
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 4
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v24
; CHECK-NEXT: addi a1, sp, 16
; CHECK-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v0
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v0, v8
; CHECK-NEXT: addi a1, sp, 16
; CHECK-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v0, v24, v8
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v16, v0
; CHECK-NEXT: vmv8r.v v8, v16
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 5
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 32 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 32 x bfloat> %elt.head, <vscale x 32 x bfloat> poison, <vscale x 32 x i32> zeroinitializer
%v = call <vscale x 32 x bfloat> @llvm.vp.fma.nxv32bf16(<vscale x 32 x bfloat> %va, <vscale x 32 x bfloat> %vb, <vscale x 32 x bfloat> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x bfloat> %v
}
define <vscale x 32 x bfloat> @vfma_vf_nxv32bf16_unmasked_commute(<vscale x 32 x bfloat> %va, bfloat %b, <vscale x 32 x bfloat> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv32bf16_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 5
; CHECK-NEXT: sub sp, sp, a1
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; CHECK-NEXT: fmv.x.h a1, fa0
; CHECK-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; CHECK-NEXT: vmv.v.x v24, a1
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a1, a2, 1
; CHECK-NEXT: sub a3, a0, a1
; CHECK-NEXT: sltu a4, a0, a3
; CHECK-NEXT: addi a4, a4, -1
; CHECK-NEXT: and a3, a4, a3
; CHECK-NEXT: srli a2, a2, 2
; CHECK-NEXT: vmset.m v7
; CHECK-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; CHECK-NEXT: vslidedown.vx v0, v7, a2
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 4
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v20
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v12
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: mv a4, a2
; CHECK-NEXT: slli a2, a2, 1
; CHECK-NEXT: add a2, a2, a4
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v8, v28
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v8, v24, v0.t
; CHECK-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v12, v16
; CHECK-NEXT: bltu a0, a1, .LBB35_2
; CHECK-NEXT: # %bb.1:
; CHECK-NEXT: mv a0, a1
; CHECK-NEXT: .LBB35_2:
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 4
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v16, v24
; CHECK-NEXT: addi a1, sp, 16
; CHECK-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v24, v0
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: mv a2, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a2
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vfwcvtbf16.f.f.v v0, v16
; CHECK-NEXT: addi a1, sp, 16
; CHECK-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v0, v24, v16
; CHECK-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; CHECK-NEXT: vfncvtbf16.f.f.w v8, v0
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 5
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 32 x bfloat> poison, bfloat %b, i32 0
%vb = shufflevector <vscale x 32 x bfloat> %elt.head, <vscale x 32 x bfloat> poison, <vscale x 32 x i32> zeroinitializer
%v = call <vscale x 32 x bfloat> @llvm.vp.fma.nxv32bf16(<vscale x 32 x bfloat> %vb, <vscale x 32 x bfloat> %va, <vscale x 32 x bfloat> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x bfloat> %v
}
declare <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x i1>, i32)
define <vscale x 1 x half> @vfma_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv1f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmadd.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv1r.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv1f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v10, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfma_vv_nxv1f16_unmasked(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv1f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv1f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfma_vf_nxv1f16(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv1f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv1f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfma_vf_nxv1f16_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv1f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv1f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v8, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %va, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfma_vf_nxv1f16_unmasked(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv1f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv1f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfma_vf_nxv1f16_unmasked_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv1f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv1f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %va, <vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
declare <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x i1>, i32)
define <vscale x 2 x half> @vfma_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv2f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmadd.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv1r.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv2f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v10, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfma_vv_nxv2f16_unmasked(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv2f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv2f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfma_vf_nxv2f16(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv2f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv2f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfma_vf_nxv2f16_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv2f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv2f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v8, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %va, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfma_vf_nxv2f16_unmasked(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv2f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv2f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfma_vf_nxv2f16_unmasked_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv2f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv2f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %va, <vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
declare <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x i1>, i32)
define <vscale x 4 x half> @vfma_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv4f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmadd.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv.v.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv4f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v10, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfma_vv_nxv4f16_unmasked(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv4f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv4f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v10, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfma_vf_nxv4f16(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv4f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv4f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v14, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfma_vf_nxv4f16_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv4f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv4f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v8, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %va, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfma_vf_nxv4f16_unmasked(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv4f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv4f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v14, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfma_vf_nxv4f16_unmasked_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv4f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv4f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v14, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %va, <vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
declare <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x i1>, i32)
define <vscale x 8 x half> @vfma_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv8f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmadd.vv v10, v8, v12, v0.t
; ZVFH-NEXT: vmv.v.v v8, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv8f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v12, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfma_vv_nxv8f16_unmasked(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv8f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmadd.vv v8, v10, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv8f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v12, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfma_vf_nxv8f16(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv8f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv8f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v20, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfma_vf_nxv8f16_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv8f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv8f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v8, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %va, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfma_vf_nxv8f16_unmasked(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv8f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv8f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v20, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfma_vf_nxv8f16_unmasked_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv8f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv8f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v20, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %va, <vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
declare <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x i1>, i32)
define <vscale x 16 x half> @vfma_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv16f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmadd.vv v12, v8, v16, v0.t
; ZVFH-NEXT: vmv.v.v v8, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv16f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 8 * vlenb
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfma_vv_nxv16f16_unmasked(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv16f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmadd.vv v8, v12, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv16f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v16, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: ret
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfma_vf_nxv16f16(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv16f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv16f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 8 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v4, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfma_vf_nxv16f16_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv16f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv16f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v4, a1
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %va, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfma_vf_nxv16f16_unmasked(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv16f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv16f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 4 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs4r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: vl4r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v0, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfma_vf_nxv16f16_unmasked_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv16f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv16f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 4 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs4r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: vl4r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v0, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %va, <vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
declare <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x i1>, i32)
define <vscale x 32 x half> @vfma_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv32f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFH-NEXT: vmv.v.v v8, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv32f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a3, a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 2
; ZVFHMIN-NEXT: add a2, a2, a3
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x29, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 41 * vlenb
; ZVFHMIN-NEXT: vmv1r.v v24, v0
; ZVFHMIN-NEXT: vl8re16.v v0, (a0)
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a0, a2, 1
; ZVFHMIN-NEXT: sub a3, a1, a0
; ZVFHMIN-NEXT: sltu a4, a1, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs1r.v v24, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v24, v24, a2
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs1r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a4, a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vmv8r.v v8, v16
; ZVFHMIN-NEXT: vmv4r.v v24, v8
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a4, a2, 5
; ZVFHMIN-NEXT: add a2, a4, a2
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a4, a2, 4
; ZVFHMIN-NEXT: add a2, a4, a2
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v0, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl1r.v v0, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a1, a0, .LBB66_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: .LBB66_2:
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a2, a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a2, a0, 5
; ZVFHMIN-NEXT: add a0, a2, a0
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a2, a0, 4
; ZVFHMIN-NEXT: add a0, a2, a0
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a2, a0, 5
; ZVFHMIN-NEXT: add a0, a2, a0
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 4
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl1r.v v0, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a2, a0, 5
; ZVFHMIN-NEXT: add a0, a2, a0
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vmv.v.v v16, v8
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a1, a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add a0, a0, a1
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfma_vv_nxv32f16_unmasked(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv32f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfmadd.vv v8, v16, v24
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv32f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 5
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: vmv8r.v v24, v16
; ZVFHMIN-NEXT: vl8re16.v v16, (a0)
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a0, a2, 1
; ZVFHMIN-NEXT: sub a3, a1, a0
; ZVFHMIN-NEXT: sltu a4, a1, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v28
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v28
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: bltu a1, a0, .LBB67_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: .LBB67_2:
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 4
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v16
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmacc.vv v0, v16, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfma_vf_nxv32f16(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv32f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv32f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a2, a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x29, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 41 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 5
; ZVFHMIN-NEXT: add a1, a2, a1
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs1r.v v0, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a2
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a4, a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a4, a2, 4
; ZVFHMIN-NEXT: add a2, a4, a2
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a4, a2, 5
; ZVFHMIN-NEXT: add a2, a4, a2
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v28
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v20, v8
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a0, a1, .LBB68_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB68_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a2, a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 4
; ZVFHMIN-NEXT: add a1, a2, a1
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a2, a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 5
; ZVFHMIN-NEXT: add a1, a2, a1
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl1r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a2, a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vmv.v.v v16, v8
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a1, a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add a0, a0, a1
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfma_vf_nxv32f16_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv32f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv32f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a2, a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x29, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 41 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 5
; ZVFHMIN-NEXT: add a1, a2, a1
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs1r.v v0, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a2
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a4, a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a4, a2, 4
; ZVFHMIN-NEXT: add a2, a4, a2
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a4, a2, 5
; ZVFHMIN-NEXT: add a2, a4, a2
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v28
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a0, a1, .LBB69_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB69_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a2, a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 4
; ZVFHMIN-NEXT: add a1, a2, a1
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 5
; ZVFHMIN-NEXT: add a1, a2, a1
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a2, a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl1r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a2, a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v24, v16, v0.t
; ZVFHMIN-NEXT: vmv.v.v v16, v8
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a1, a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add a0, a0, a1
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %va, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfma_vf_nxv32f16_unmasked(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv32f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv32f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v28
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v20, v8
; ZVFHMIN-NEXT: bltu a0, a1, .LBB70_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB70_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v24, v8
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v16, v0
; ZVFHMIN-NEXT: vmv8r.v v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfma_vf_nxv32f16_unmasked_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vf_nxv32f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfmadd.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vf_nxv32f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v28
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: bltu a0, a1, .LBB71_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB71_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v16
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v24, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %va, <vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
declare <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x i1>, i32)
define <vscale x 1 x float> @vfma_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv1r.v v8, v9
; CHECK-NEXT: ret
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfma_vv_nxv1f32_unmasked(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv1f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfma_vf_nxv1f32(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfma_vf_nxv1f32_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %va, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfma_vf_nxv1f32_unmasked(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfma_vf_nxv1f32_unmasked_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %va, <vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
declare <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x i1>, i32)
define <vscale x 2 x float> @vfma_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfma_vv_nxv2f32_unmasked(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv2f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfma_vf_nxv2f32(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfma_vf_nxv2f32_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %va, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfma_vf_nxv2f32_unmasked(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfma_vf_nxv2f32_unmasked_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %va, <vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
declare <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x i1>, i32)
define <vscale x 4 x float> @vfma_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vv v10, v8, v12, v0.t
; CHECK-NEXT: vmv.v.v v8, v10
; CHECK-NEXT: ret
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfma_vv_nxv4f32_unmasked(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv4f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v10, v12
; CHECK-NEXT: ret
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfma_vf_nxv4f32(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfma_vf_nxv4f32_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %va, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfma_vf_nxv4f32_unmasked(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfma_vf_nxv4f32_unmasked_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %va, <vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
declare <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x i1>, i32)
define <vscale x 8 x float> @vfma_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv8f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v8, v16, v0.t
; CHECK-NEXT: vmv.v.v v8, v12
; CHECK-NEXT: ret
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfma_vv_nxv8f32_unmasked(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv8f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v12, v16
; CHECK-NEXT: ret
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfma_vf_nxv8f32(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfma_vf_nxv8f32_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %va, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfma_vf_nxv8f32_unmasked(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfma_vf_nxv8f32_unmasked_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %va, <vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
declare <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x i1>, i32)
define <vscale x 16 x float> @vfma_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv16f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v8, v24, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: ret
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfma_vv_nxv16f32_unmasked(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv16f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v16, v24
; CHECK-NEXT: ret
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfma_vf_nxv16f32(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv16f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfma_vf_nxv16f32_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv16f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %va, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfma_vf_nxv16f32_unmasked(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv16f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfma_vf_nxv16f32_unmasked_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv16f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %va, <vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
declare <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x i1>, i32)
define <vscale x 1 x double> @vfma_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmadd.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfma_vv_nxv1f64_unmasked(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv1f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfma_vf_nxv1f64(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfma_vf_nxv1f64_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %va, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfma_vf_nxv1f64_unmasked(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfma_vf_nxv1f64_unmasked_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv1f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %va, <vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
declare <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x i1>, i32)
define <vscale x 2 x double> @vfma_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmadd.vv v10, v8, v12, v0.t
; CHECK-NEXT: vmv.v.v v8, v10
; CHECK-NEXT: ret
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfma_vv_nxv2f64_unmasked(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv2f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v10, v12
; CHECK-NEXT: ret
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfma_vf_nxv2f64(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfma_vf_nxv2f64_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %va, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfma_vf_nxv2f64_unmasked(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfma_vf_nxv2f64_unmasked_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv2f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %va, <vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
declare <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x i1>, i32)
define <vscale x 4 x double> @vfma_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmadd.vv v12, v8, v16, v0.t
; CHECK-NEXT: vmv.v.v v8, v12
; CHECK-NEXT: ret
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfma_vv_nxv4f64_unmasked(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv4f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v12, v16
; CHECK-NEXT: ret
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfma_vf_nxv4f64(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfma_vf_nxv4f64_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %va, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfma_vf_nxv4f64_unmasked(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfma_vf_nxv4f64_unmasked_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv4f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %va, <vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
declare <vscale x 7 x double> @llvm.vp.fma.nxv7f64(<vscale x 7 x double>, <vscale x 7 x double>, <vscale x 7 x double>, <vscale x 7 x i1>, i32)
define <vscale x 7 x double> @vfma_vv_nxv7f64(<vscale x 7 x double> %va, <vscale x 7 x double> %b, <vscale x 7 x double> %c, <vscale x 7 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv7f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v8, v24, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: ret
%v = call <vscale x 7 x double> @llvm.vp.fma.nxv7f64(<vscale x 7 x double> %va, <vscale x 7 x double> %b, <vscale x 7 x double> %c, <vscale x 7 x i1> %m, i32 %evl)
ret <vscale x 7 x double> %v
}
define <vscale x 7 x double> @vfma_vv_nxv7f64_unmasked(<vscale x 7 x double> %va, <vscale x 7 x double> %b, <vscale x 7 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv7f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v16, v24
; CHECK-NEXT: ret
%v = call <vscale x 7 x double> @llvm.vp.fma.nxv7f64(<vscale x 7 x double> %va, <vscale x 7 x double> %b, <vscale x 7 x double> %c, <vscale x 7 x i1> splat (i1 true), i32 %evl)
ret <vscale x 7 x double> %v
}
declare <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x i1>, i32)
define <vscale x 8 x double> @vfma_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv8f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v8, v24, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: ret
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfma_vv_nxv8f64_unmasked(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv8f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v16, v24
; CHECK-NEXT: ret
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfma_vf_nxv8f64(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfma_vf_nxv8f64_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %va, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfma_vf_nxv8f64_unmasked(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfma_vf_nxv8f64_unmasked_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vf_nxv8f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %va, <vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
declare <vscale x 16 x double> @llvm.vp.fma.nxv16f64(<vscale x 16 x double>, <vscale x 16 x double>, <vscale x 16 x double>, <vscale x 16 x i1>, i32)
define <vscale x 16 x double> @vfma_vv_nxv16f64(<vscale x 16 x double> %va, <vscale x 16 x double> %b, <vscale x 16 x double> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv16f64:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: mv a3, a1
; CHECK-NEXT: slli a1, a1, 2
; CHECK-NEXT: add a1, a1, a3
; CHECK-NEXT: sub sp, sp, a1
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x28, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 40 * vlenb
; CHECK-NEXT: vmv1r.v v7, v0
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: mv a3, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a3
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 5
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: srli a3, a1, 3
; CHECK-NEXT: slli a5, a1, 3
; CHECK-NEXT: add a6, a2, a5
; CHECK-NEXT: vl8re64.v v8, (a6)
; CHECK-NEXT: csrr a6, vlenb
; CHECK-NEXT: slli a6, a6, 4
; CHECK-NEXT: add a6, sp, a6
; CHECK-NEXT: addi a6, a6, 16
; CHECK-NEXT: vs8r.v v8, (a6) # Unknown-size Folded Spill
; CHECK-NEXT: add a5, a0, a5
; CHECK-NEXT: sub a6, a4, a1
; CHECK-NEXT: sltu a7, a4, a6
; CHECK-NEXT: addi a7, a7, -1
; CHECK-NEXT: vl8re64.v v8, (a5)
; CHECK-NEXT: vl8re64.v v16, (a2)
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vl8re64.v v16, (a0)
; CHECK-NEXT: addi a0, sp, 16
; CHECK-NEXT: vs8r.v v16, (a0) # Unknown-size Folded Spill
; CHECK-NEXT: vsetvli a0, zero, e8, mf4, ta, ma
; CHECK-NEXT: vslidedown.vx v0, v0, a3
; CHECK-NEXT: and a0, a7, a6
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 3
; CHECK-NEXT: mv a3, a2
; CHECK-NEXT: slli a2, a2, 1
; CHECK-NEXT: add a2, a2, a3
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: csrr a2, vlenb
; CHECK-NEXT: slli a2, a2, 4
; CHECK-NEXT: add a2, sp, a2
; CHECK-NEXT: addi a2, a2, 16
; CHECK-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v8, v24, v16, v0.t
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: mv a2, a0
; CHECK-NEXT: slli a0, a0, 1
; CHECK-NEXT: add a0, a0, a2
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vs8r.v v8, (a0) # Unknown-size Folded Spill
; CHECK-NEXT: bltu a4, a1, .LBB128_2
; CHECK-NEXT: # %bb.1:
; CHECK-NEXT: mv a4, a1
; CHECK-NEXT: .LBB128_2:
; CHECK-NEXT: vmv1r.v v0, v7
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 5
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: addi a0, sp, 16
; CHECK-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a4, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v24, v8, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: mv a1, a0
; CHECK-NEXT: slli a0, a0, 1
; CHECK-NEXT: add a0, a0, a1
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: mv a1, a0
; CHECK-NEXT: slli a0, a0, 2
; CHECK-NEXT: add a0, a0, a1
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%v = call <vscale x 16 x double> @llvm.vp.fma.nxv16f64(<vscale x 16 x double> %va, <vscale x 16 x double> %b, <vscale x 16 x double> %c, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x double> %v
}
define <vscale x 16 x double> @vfma_vv_nxv16f64_unmasked(<vscale x 16 x double> %va, <vscale x 16 x double> %b, <vscale x 16 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfma_vv_nxv16f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: addi sp, sp, -16
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: mv a3, a1
; CHECK-NEXT: slli a1, a1, 1
; CHECK-NEXT: add a1, a1, a3
; CHECK-NEXT: sub sp, sp, a1
; CHECK-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x18, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 24 * vlenb
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 3
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a1, a1, 4
; CHECK-NEXT: add a1, sp, a1
; CHECK-NEXT: addi a1, a1, 16
; CHECK-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; CHECK-NEXT: csrr a1, vlenb
; CHECK-NEXT: slli a3, a1, 3
; CHECK-NEXT: add a5, a2, a3
; CHECK-NEXT: vl8re64.v v24, (a5)
; CHECK-NEXT: add a3, a0, a3
; CHECK-NEXT: vl8re64.v v16, (a3)
; CHECK-NEXT: sub a3, a4, a1
; CHECK-NEXT: sltu a5, a4, a3
; CHECK-NEXT: vl8re64.v v8, (a2)
; CHECK-NEXT: addi a2, sp, 16
; CHECK-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; CHECK-NEXT: vl8re64.v v0, (a0)
; CHECK-NEXT: addi a5, a5, -1
; CHECK-NEXT: and a3, a5, a3
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a3, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v16, v8, v24
; CHECK-NEXT: bltu a4, a1, .LBB129_2
; CHECK-NEXT: # %bb.1:
; CHECK-NEXT: mv a4, a1
; CHECK-NEXT: .LBB129_2:
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 4
; CHECK-NEXT: add a0, sp, a0
; CHECK-NEXT: addi a0, a0, 16
; CHECK-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: addi a0, sp, 16
; CHECK-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-NEXT: vsetvli zero, a4, e64, m8, ta, ma
; CHECK-NEXT: vfmadd.vv v0, v24, v8
; CHECK-NEXT: vmv.v.v v8, v0
; CHECK-NEXT: csrr a0, vlenb
; CHECK-NEXT: slli a0, a0, 3
; CHECK-NEXT: mv a1, a0
; CHECK-NEXT: slli a0, a0, 1
; CHECK-NEXT: add a0, a0, a1
; CHECK-NEXT: add sp, sp, a0
; CHECK-NEXT: addi sp, sp, 16
; CHECK-NEXT: ret
%v = call <vscale x 16 x double> @llvm.vp.fma.nxv16f64(<vscale x 16 x double> %va, <vscale x 16 x double> %b, <vscale x 16 x double> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x double> %v
}
declare <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half>, <vscale x 1 x i1>, i32)
define <vscale x 1 x half> @vfmsub_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv1f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmsub.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv1r.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv1f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v10, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfmsub_vv_nxv1f16_unmasked(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv1f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmsub.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv1f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfmsub_vf_nxv1f16(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv1f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv1f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfmsub_vf_nxv1f16_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv1f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv1f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v8, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %va, <vscale x 1 x half> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfmsub_vf_nxv1f16_unmasked(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv1f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv1f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfmsub_vf_nxv1f16_unmasked_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv1f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv1f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %va, <vscale x 1 x half> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv1f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv1r.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv1f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %negb, <vscale x 1 x half> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vv_nxv1f16_commuted(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv1f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv1f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negb, <vscale x 1 x half> %va, <vscale x 1 x half> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vv_nxv1f16_unmasked(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv1f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv1f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %negb, <vscale x 1 x half> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vv_nxv1f16_unmasked_commuted(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv1f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv1f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negb, <vscale x 1 x half> %va, <vscale x 1 x half> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vf_nxv1f16(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv1f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv1f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negva, <vscale x 1 x half> %vb, <vscale x 1 x half> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vf_nxv1f16_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv1f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv1f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v8, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %negva, <vscale x 1 x half> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vf_nxv1f16_unmasked(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv1f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv1f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negva, <vscale x 1 x half> %vb, <vscale x 1 x half> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vf_nxv1f16_unmasked_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv1f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv1f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %negva, <vscale x 1 x half> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vf_nxv1f16_neg_splat(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv1f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv1f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %negvb, <vscale x 1 x half> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vf_nxv1f16_neg_splat_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv1f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv1f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negvb, <vscale x 1 x half> %va, <vscale x 1 x half> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vf_nxv1f16_neg_splat_unmasked(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv1f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv1f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %negvb, <vscale x 1 x half> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmadd_vf_nxv1f16_neg_splat_unmasked_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv1f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv1f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negvb, <vscale x 1 x half> %va, <vscale x 1 x half> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv1f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv1r.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv1f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %negb, <vscale x 1 x half> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vv_nxv1f16_commuted(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv1f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv1f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negb, <vscale x 1 x half> %va, <vscale x 1 x half> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vv_nxv1f16_unmasked(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv1f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv1f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %negb, <vscale x 1 x half> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vv_nxv1f16_unmasked_commuted(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv1f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv1f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negb, <vscale x 1 x half> %va, <vscale x 1 x half> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vf_nxv1f16(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv1f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv1f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negva, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vf_nxv1f16_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv1f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv1f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v11, v9, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v11
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %negva, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vf_nxv1f16_unmasked(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv1f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv1f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v8
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negva, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vf_nxv1f16_unmasked_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv1f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv1f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v8
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %negva, <vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vf_nxv1f16_neg_splat(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv1f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv1f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v11, v9, v10, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v11
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %negvb, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vf_nxv1f16_neg_splat_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv1f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv1f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v10, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negvb, <vscale x 1 x half> %va, <vscale x 1 x half> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vf_nxv1f16_neg_splat_unmasked(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv1f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv1f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v10
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %negvb, <vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
define <vscale x 1 x half> @vfnmsub_vf_nxv1f16_neg_splat_unmasked_commute(<vscale x 1 x half> %va, half %b, <vscale x 1 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv1f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv1f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v10
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 1 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %negvb, <vscale x 1 x half> %va, <vscale x 1 x half> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x half> %v
}
declare <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half>, <vscale x 2 x i1>, i32)
define <vscale x 2 x half> @vfmsub_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv2f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmsub.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv1r.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv2f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v10, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfmsub_vv_nxv2f16_unmasked(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv2f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmsub.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv2f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfmsub_vf_nxv2f16(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv2f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv2f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfmsub_vf_nxv2f16_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv2f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv2f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v8, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %va, <vscale x 2 x half> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfmsub_vf_nxv2f16_unmasked(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv2f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv2f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfmsub_vf_nxv2f16_unmasked_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv2f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv2f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %va, <vscale x 2 x half> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv2f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv1r.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv2f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %negb, <vscale x 2 x half> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vv_nxv2f16_commuted(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv2f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv2f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negb, <vscale x 2 x half> %va, <vscale x 2 x half> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vv_nxv2f16_unmasked(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv2f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv2f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %negb, <vscale x 2 x half> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vv_nxv2f16_unmasked_commuted(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv2f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv2f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negb, <vscale x 2 x half> %va, <vscale x 2 x half> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vf_nxv2f16(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv2f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv2f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negva, <vscale x 2 x half> %vb, <vscale x 2 x half> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vf_nxv2f16_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv2f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv2f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v8, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %negva, <vscale x 2 x half> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vf_nxv2f16_unmasked(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv2f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv2f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negva, <vscale x 2 x half> %vb, <vscale x 2 x half> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vf_nxv2f16_unmasked_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv2f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv2f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %negva, <vscale x 2 x half> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vf_nxv2f16_neg_splat(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv2f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv2f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %negvb, <vscale x 2 x half> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vf_nxv2f16_neg_splat_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv2f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv2f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negvb, <vscale x 2 x half> %va, <vscale x 2 x half> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vf_nxv2f16_neg_splat_unmasked(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv2f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv2f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %negvb, <vscale x 2 x half> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmadd_vf_nxv2f16_neg_splat_unmasked_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv2f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv2f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v9, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negvb, <vscale x 2 x half> %va, <vscale x 2 x half> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv2f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv1r.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv2f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v9, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %negb, <vscale x 2 x half> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vv_nxv2f16_commuted(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv2f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv2f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negb, <vscale x 2 x half> %va, <vscale x 2 x half> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vv_nxv2f16_unmasked(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv2f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv2f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %negb, <vscale x 2 x half> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vv_nxv2f16_unmasked_commuted(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv2f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv2f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v10, v11
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negb, <vscale x 2 x half> %va, <vscale x 2 x half> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vf_nxv2f16(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv2f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv2f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negva, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vf_nxv2f16_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv2f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv2f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v11, v9, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v11
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %negva, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vf_nxv2f16_unmasked(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv2f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv2f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v8
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negva, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vf_nxv2f16_unmasked_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv2f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv2f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v8
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %negva, <vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vf_nxv2f16_neg_splat(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv2f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv2f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v11, v9, v10, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v11
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %negvb, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vf_nxv2f16_neg_splat_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv2f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv2f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v10, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negvb, <vscale x 2 x half> %va, <vscale x 2 x half> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vf_nxv2f16_neg_splat_unmasked(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv2f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv2f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v10
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %negvb, <vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
define <vscale x 2 x half> @vfnmsub_vf_nxv2f16_neg_splat_unmasked_commute(<vscale x 2 x half> %va, half %b, <vscale x 2 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv2f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv2f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v9, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v9, v11, v10
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v9
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 2 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x half> @llvm.vp.fneg.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x half> @llvm.vp.fma.nxv2f16(<vscale x 2 x half> %negvb, <vscale x 2 x half> %va, <vscale x 2 x half> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x half> %v
}
declare <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half>, <vscale x 4 x i1>, i32)
define <vscale x 4 x half> @vfmsub_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv4f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmsub.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv.v.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv4f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v10, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfmsub_vv_nxv4f16_unmasked(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv4f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmsub.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv4f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v10, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfmsub_vf_nxv4f16(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv4f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv4f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v14, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfmsub_vf_nxv4f16_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv4f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv4f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v8, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %va, <vscale x 4 x half> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfmsub_vf_nxv4f16_unmasked(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv4f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv4f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v14, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfmsub_vf_nxv4f16_unmasked_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv4f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv4f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v14, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %va, <vscale x 4 x half> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv4f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv.v.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv4f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v14, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %negb, <vscale x 4 x half> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vv_nxv4f16_commuted(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv4f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv4f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v10, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negb, <vscale x 4 x half> %va, <vscale x 4 x half> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vv_nxv4f16_unmasked(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv4f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv4f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v10, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %negb, <vscale x 4 x half> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vv_nxv4f16_unmasked_commuted(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv4f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv4f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v10, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negb, <vscale x 4 x half> %va, <vscale x 4 x half> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vf_nxv4f16(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv4f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv4f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v14, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negva, <vscale x 4 x half> %vb, <vscale x 4 x half> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vf_nxv4f16_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv4f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv4f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v8, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %negva, <vscale x 4 x half> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vf_nxv4f16_unmasked(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv4f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv4f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v14, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negva, <vscale x 4 x half> %vb, <vscale x 4 x half> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vf_nxv4f16_unmasked_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv4f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv4f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v14, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %negva, <vscale x 4 x half> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vf_nxv4f16_neg_splat(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv4f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv4f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v10, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %negvb, <vscale x 4 x half> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vf_nxv4f16_neg_splat_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv4f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv4f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v14, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negvb, <vscale x 4 x half> %va, <vscale x 4 x half> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vf_nxv4f16_neg_splat_unmasked(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv4f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv4f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v14, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %negvb, <vscale x 4 x half> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmadd_vf_nxv4f16_neg_splat_unmasked_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv4f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv4f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v14, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negvb, <vscale x 4 x half> %va, <vscale x 4 x half> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv4f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv.v.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv4f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v10, v14, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v10
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %negb, <vscale x 4 x half> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vv_nxv4f16_commuted(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv4f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv4f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v10, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negb, <vscale x 4 x half> %va, <vscale x 4 x half> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vv_nxv4f16_unmasked(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv4f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv4f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v10, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %negb, <vscale x 4 x half> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vv_nxv4f16_unmasked_commuted(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv4f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v9, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv4f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v9, v9, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v10, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negb, <vscale x 4 x half> %va, <vscale x 4 x half> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vf_nxv4f16(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv4f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv4f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v12, v14, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negva, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vf_nxv4f16_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv4f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv4f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v8, v14, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %negva, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vf_nxv4f16_unmasked(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv4f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv4f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v12, v14
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negva, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vf_nxv4f16_unmasked_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv4f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv4f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v12, v14
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %negva, <vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vf_nxv4f16_neg_splat(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv4f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv4f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v14, v10, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %negvb, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vf_nxv4f16_neg_splat_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv4f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv4f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v12, v10, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negvb, <vscale x 4 x half> %va, <vscale x 4 x half> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vf_nxv4f16_neg_splat_unmasked(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv4f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv4f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v12, v10
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %negvb, <vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
define <vscale x 4 x half> @vfnmsub_vf_nxv4f16_neg_splat_unmasked_commute(<vscale x 4 x half> %va, half %b, <vscale x 4 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv4f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv4f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v10, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m1, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
; ZVFHMIN-NEXT: vfwcvt.f.f.v v14, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v14, v12, v10
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m1, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v14
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 4 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x half> @llvm.vp.fneg.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x half> @llvm.vp.fma.nxv4f16(<vscale x 4 x half> %negvb, <vscale x 4 x half> %va, <vscale x 4 x half> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x half> %v
}
declare <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half>, <vscale x 8 x i1>, i32)
define <vscale x 8 x half> @vfmsub_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv8f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmsub.vv v10, v8, v12, v0.t
; ZVFH-NEXT: vmv.v.v v8, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv8f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v12, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfmsub_vv_nxv8f16_unmasked(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv8f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmsub.vv v8, v10, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv8f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v10
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v12, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfmsub_vf_nxv8f16(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv8f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv8f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v20, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfmsub_vf_nxv8f16_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv8f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv8f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v8, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %va, <vscale x 8 x half> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfmsub_vf_nxv8f16_unmasked(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv8f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv8f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v20, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfmsub_vf_nxv8f16_unmasked_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv8f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv8f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v20, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %va, <vscale x 8 x half> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv8f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v10, v8, v12, v0.t
; ZVFH-NEXT: vmv.v.v v8, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv8f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v20, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %negb, <vscale x 8 x half> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vv_nxv8f16_commuted(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv8f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v10, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv8f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v12, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negb, <vscale x 8 x half> %va, <vscale x 8 x half> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vv_nxv8f16_unmasked(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv8f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v10, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv8f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v12, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %negb, <vscale x 8 x half> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vv_nxv8f16_unmasked_commuted(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv8f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v10, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv8f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v12, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negb, <vscale x 8 x half> %va, <vscale x 8 x half> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vf_nxv8f16(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv8f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv8f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v20, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negva, <vscale x 8 x half> %vb, <vscale x 8 x half> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vf_nxv8f16_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv8f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv8f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v8, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %negva, <vscale x 8 x half> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vf_nxv8f16_unmasked(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv8f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv8f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v20, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negva, <vscale x 8 x half> %vb, <vscale x 8 x half> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vf_nxv8f16_unmasked_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv8f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv8f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v20, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %negva, <vscale x 8 x half> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vf_nxv8f16_neg_splat(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv8f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv8f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v12, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %negvb, <vscale x 8 x half> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vf_nxv8f16_neg_splat_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv8f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv8f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v20, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negvb, <vscale x 8 x half> %va, <vscale x 8 x half> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vf_nxv8f16_neg_splat_unmasked(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv8f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv8f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v20, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %negvb, <vscale x 8 x half> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmadd_vf_nxv8f16_neg_splat_unmasked_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv8f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv8f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v20, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negvb, <vscale x 8 x half> %va, <vscale x 8 x half> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv8f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v10, v8, v12, v0.t
; ZVFH-NEXT: vmv.v.v v8, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv8f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v20, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %negb, <vscale x 8 x half> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vv_nxv8f16_commuted(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv8f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v10, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv8f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v12, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negb, <vscale x 8 x half> %va, <vscale x 8 x half> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vv_nxv8f16_unmasked(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv8f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v10, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv8f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v12, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %negb, <vscale x 8 x half> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vv_nxv8f16_unmasked_commuted(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv8f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v10, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv8f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v10, v10, a1
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v12, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negb, <vscale x 8 x half> %va, <vscale x 8 x half> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vf_nxv8f16(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv8f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv8f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v20, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negva, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vf_nxv8f16_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv8f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv8f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v20, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %negva, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vf_nxv8f16_unmasked(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv8f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv8f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v20
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negva, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vf_nxv8f16_unmasked_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv8f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv8f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v20
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %negva, <vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vf_nxv8f16_neg_splat(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv8f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv8f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v20, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %negvb, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vf_nxv8f16_neg_splat_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv8f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv8f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v16, v12, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negvb, <vscale x 8 x half> %va, <vscale x 8 x half> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vf_nxv8f16_neg_splat_unmasked(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv8f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv8f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v16, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %negvb, <vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
define <vscale x 8 x half> @vfnmsub_vf_nxv8f16_neg_splat_unmasked_commute(<vscale x 8 x half> %va, half %b, <vscale x 8 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv8f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v10
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv8f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v12, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m2, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v20, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v20, v16, v12
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m2, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v20
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 8 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x half> @llvm.vp.fneg.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x half> @llvm.vp.fma.nxv8f16(<vscale x 8 x half> %negvb, <vscale x 8 x half> %va, <vscale x 8 x half> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x half> %v
}
declare <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half>, <vscale x 16 x i1>, i32)
define <vscale x 16 x half> @vfmsub_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv16f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmsub.vv v12, v8, v16, v0.t
; ZVFH-NEXT: vmv.v.v v8, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv16f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: vmv4r.v v20, v8
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %negc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfmsub_vv_nxv16f16_unmasked(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv16f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmsub.vv v8, v12, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv16f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v12
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v16, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %negc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfmsub_vf_nxv16f16(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv16f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv16f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vmv4r.v v16, v8
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v4, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfmsub_vf_nxv16f16_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv16f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv16f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v4, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %va, <vscale x 16 x half> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfmsub_vf_nxv16f16_unmasked(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv16f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv16f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 4 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs4r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vl4r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v0, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfmsub_vf_nxv16f16_unmasked_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv16f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv16f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 4 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs4r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vl4r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v0, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %va, <vscale x 16 x half> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv16f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v12, v8, v16, v0.t
; ZVFH-NEXT: vmv.v.v v8, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv16f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vmv4r.v v4, v8
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
%negc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %negb, <vscale x 16 x half> %negc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vv_nxv16f16_commuted(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv16f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v12, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv16f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 8 * vlenb
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
%negc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negb, <vscale x 16 x half> %va, <vscale x 16 x half> %negc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vv_nxv16f16_unmasked(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv16f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v12, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv16f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v16, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %b, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %negb, <vscale x 16 x half> %negc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vv_nxv16f16_unmasked_commuted(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv16f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v12, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv16f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v16, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %b, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negb, <vscale x 16 x half> %va, <vscale x 16 x half> %negc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vf_nxv16f16(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv16f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv16f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 8 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v4, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x i1> %m, i32 %evl)
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negva, <vscale x 16 x half> %vb, <vscale x 16 x half> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vf_nxv16f16_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv16f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv16f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v4, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x i1> %m, i32 %evl)
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %negva, <vscale x 16 x half> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vf_nxv16f16_unmasked(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv16f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv16f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 4 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs4r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vl4r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v0, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negva, <vscale x 16 x half> %vb, <vscale x 16 x half> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vf_nxv16f16_unmasked_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv16f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv16f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 4 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs4r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vl4r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v0, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %negva, <vscale x 16 x half> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vf_nxv16f16_neg_splat(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv16f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv16f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 4 * vlenb
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs4r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v4, v16, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vl4r.v v4, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %negvb, <vscale x 16 x half> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vf_nxv16f16_neg_splat_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv16f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv16f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vmv4r.v v4, v8
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negvb, <vscale x 16 x half> %va, <vscale x 16 x half> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vf_nxv16f16_neg_splat_unmasked(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv16f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv16f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v0, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %negvb, <vscale x 16 x half> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmadd_vf_nxv16f16_neg_splat_unmasked_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv16f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv16f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v0, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negvb, <vscale x 16 x half> %va, <vscale x 16 x half> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv16f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v12, v8, v16, v0.t
; ZVFH-NEXT: vmv.v.v v8, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv16f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vmv4r.v v4, v8
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
%negc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %negb, <vscale x 16 x half> %negc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vv_nxv16f16_commuted(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv16f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v12, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv16f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x08, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 8 * vlenb
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
%negc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %c, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negb, <vscale x 16 x half> %va, <vscale x 16 x half> %negc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vv_nxv16f16_unmasked(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv16f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v12, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv16f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v16, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %b, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %negb, <vscale x 16 x half> %negc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vv_nxv16f16_unmasked_commuted(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv16f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v12, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv16f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v12, v12, a1
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v16, v24
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %b, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negb, <vscale x 16 x half> %va, <vscale x 16 x half> %negc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vf_nxv16f16(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv16f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv16f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vmv4r.v v16, v12
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v4, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negva, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vf_nxv16f16_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv16f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv16f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v4, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v8, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %negva, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vf_nxv16f16_unmasked(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv16f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv16f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 4 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs4r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v12
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vl4r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v0
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negva, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vf_nxv16f16_unmasked_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv16f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv16f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 4 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs4r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v12
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vl4r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v0
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %negva, <vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vf_nxv16f16_neg_splat(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv16f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv16f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vmv4r.v v4, v8
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v8, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %negvb, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vf_nxv16f16_neg_splat_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv16f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv16f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vmv4r.v v20, v12
; ZVFHMIN-NEXT: vmv4r.v v4, v8
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negvb, <vscale x 16 x half> %va, <vscale x 16 x half> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vf_nxv16f16_neg_splat_unmasked(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv16f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv16f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v24, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %negvb, <vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
define <vscale x 16 x half> @vfnmsub_vf_nxv16f16_neg_splat_unmasked_commute(<vscale x 16 x half> %va, half %b, <vscale x 16 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv16f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v12
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv16f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v16, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v24, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 16 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x half> @llvm.vp.fneg.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x half> @llvm.vp.fma.nxv16f16(<vscale x 16 x half> %negvb, <vscale x 16 x half> %va, <vscale x 16 x half> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x half> %v
}
declare <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half>, <vscale x 32 x i1>, i32)
define <vscale x 32 x half> @vfmsub_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv32f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfmsub.vv v16, v8, v24, v0.t
; ZVFH-NEXT: vmv.v.v v8, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv32f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a3, a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 2
; ZVFHMIN-NEXT: add a2, a2, a3
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x29, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 41 * vlenb
; ZVFHMIN-NEXT: vl8re16.v v24, (a0)
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a2, a0, 5
; ZVFHMIN-NEXT: add a0, a2, a0
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a0, 8
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs1r.v v0, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v24, a0, v0.t
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a2, a0, 1
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: mv a4, a3
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a4, a4, a3
; ZVFHMIN-NEXT: slli a3, a3, 1
; ZVFHMIN-NEXT: add a3, a3, a4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a4, a3, 4
; ZVFHMIN-NEXT: add a3, a4, a3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: mv a3, a1
; ZVFHMIN-NEXT: bltu a1, a2, .LBB280_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB280_2:
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a5, a4, 5
; ZVFHMIN-NEXT: add a4, a5, a4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v0
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl1r.v v0, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a5, a4, 4
; ZVFHMIN-NEXT: add a4, a5, a4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v8, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: mv a4, a3
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a4, a4, a3
; ZVFHMIN-NEXT: slli a3, a3, 1
; ZVFHMIN-NEXT: add a3, a3, a4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a4, a3, 4
; ZVFHMIN-NEXT: add a3, a4, a3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: sub a2, a1, a2
; ZVFHMIN-NEXT: sltu a1, a1, a2
; ZVFHMIN-NEXT: addi a1, a1, -1
; ZVFHMIN-NEXT: and a1, a1, a2
; ZVFHMIN-NEXT: srli a0, a0, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a2, a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a2, a0, 4
; ZVFHMIN-NEXT: add a0, a2, a0
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a2, a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a1, a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add a0, a0, a1
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %negc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfmsub_vv_nxv32f16_unmasked(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vv_nxv32f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfmsub.vv v8, v16, v24
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vv_nxv32f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 2
; ZVFHMIN-NEXT: add a2, a2, a3
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x28, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 40 * vlenb
; ZVFHMIN-NEXT: vl8re16.v v24, (a0)
; ZVFHMIN-NEXT: lui a0, 8
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v0, v24, a0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v0, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a0, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v24
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a0, a2, 1
; ZVFHMIN-NEXT: sub a3, a1, a0
; ZVFHMIN-NEXT: sltu a4, a1, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v24, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v4
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 5
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vmv8r.v v8, v16
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v4, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a1, a0, .LBB281_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: .LBB281_2:
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 4
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v8, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v0, v24
; ZVFHMIN-NEXT: vmv8r.v v8, v0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add a0, a0, a1
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %c, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %negc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfmsub_vf_nxv32f16(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv32f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv32f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 1
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: mv a3, a0
; ZVFHMIN-NEXT: bltu a0, a2, .LBB282_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB282_2:
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v8, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: sub a2, a0, a2
; ZVFHMIN-NEXT: sltu a0, a0, a2
; ZVFHMIN-NEXT: addi a0, a0, -1
; ZVFHMIN-NEXT: and a0, a0, a2
; ZVFHMIN-NEXT: srli a1, a1, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x half> %negvc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfmsub_vf_nxv32f16_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv32f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv32f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 1
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: mv a3, a0
; ZVFHMIN-NEXT: bltu a0, a2, .LBB283_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB283_2:
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: sub a2, a0, a2
; ZVFHMIN-NEXT: sltu a0, a0, a2
; ZVFHMIN-NEXT: addi a0, a0, -1
; ZVFHMIN-NEXT: and a0, a0, a2
; ZVFHMIN-NEXT: srli a1, a1, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a1
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v4
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v8, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %va, <vscale x 32 x half> %negvc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfmsub_vf_nxv32f16_unmasked(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv32f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv32f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a1, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v28
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v20, v8
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a0, a1, .LBB284_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB284_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v8, v0
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v16, v24
; ZVFHMIN-NEXT: vmv8r.v v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x half> %negvc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfmsub_vf_nxv32f16_unmasked_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfmsub_vf_nxv32f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfmsub.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfmsub_vf_nxv32f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: vmv8r.v v24, v16
; ZVFHMIN-NEXT: vmv8r.v v16, v8
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v8, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v24, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a1, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v28
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v20, v8
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a0, a1, .LBB285_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB285_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v8, v0
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v16, v24
; ZVFHMIN-NEXT: vmv8r.v v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %va, <vscale x 32 x half> %negvc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv32f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vv v16, v8, v24, v0.t
; ZVFH-NEXT: vmv.v.v v8, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv32f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 5
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: vl8re16.v v24, (a0)
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a0, 8
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v16, a0, v0.t
; ZVFHMIN-NEXT: vxor.vx v24, v24, a0, v0.t
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a2, a0, 1
; ZVFHMIN-NEXT: mv a3, a1
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: vsetvli a4, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a1, a2, .LBB286_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB286_2:
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v4
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: sub a2, a1, a2
; ZVFHMIN-NEXT: sltu a1, a1, a2
; ZVFHMIN-NEXT: addi a1, a1, -1
; ZVFHMIN-NEXT: and a1, a1, a2
; ZVFHMIN-NEXT: srli a0, a0, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v8, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
%negc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %negb, <vscale x 32 x half> %negc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vv_nxv32f16_commuted(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv32f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v16, v24, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv32f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 2
; ZVFHMIN-NEXT: add a2, a2, a3
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x28, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 40 * vlenb
; ZVFHMIN-NEXT: vl8re16.v v24, (a0)
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a0, 8
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v16, a0, v0.t
; ZVFHMIN-NEXT: vxor.vx v16, v24, a0, v0.t
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a2, a0, 1
; ZVFHMIN-NEXT: mv a3, a1
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: vsetvli a4, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a1, a2, .LBB287_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB287_2:
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v4
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: mv a4, a3
; ZVFHMIN-NEXT: slli a3, a3, 1
; ZVFHMIN-NEXT: add a3, a3, a4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v28
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: sub a2, a1, a2
; ZVFHMIN-NEXT: sltu a1, a1, a2
; ZVFHMIN-NEXT: addi a1, a1, -1
; ZVFHMIN-NEXT: and a1, a1, a2
; ZVFHMIN-NEXT: srli a0, a0, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v8, v16, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add a0, a0, a1
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
%negc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negb, <vscale x 32 x half> %va, <vscale x 32 x half> %negc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vv_nxv32f16_unmasked(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv32f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v16, v24
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv32f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 5
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vl8re16.v v24, (a0)
; ZVFHMIN-NEXT: lui a0, 8
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a0
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v24, v24, a0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a0, a2, 1
; ZVFHMIN-NEXT: sub a3, a1, a0
; ZVFHMIN-NEXT: sltu a4, a1, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v28
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v28
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v28, v8
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a1, a0, .LBB288_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: .LBB288_2:
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 4
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v8, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v24, v0
; ZVFHMIN-NEXT: vmv8r.v v8, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %b, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %c, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %negb, <vscale x 32 x half> %negc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vv_nxv32f16_unmasked_commuted(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vv_nxv32f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v16, v24
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vv_nxv32f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 5
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vl8re16.v v24, (a0)
; ZVFHMIN-NEXT: lui a0, 8
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v16, a0
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v24, a0
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a0, a2, 1
; ZVFHMIN-NEXT: sub a3, a1, a0
; ZVFHMIN-NEXT: sltu a4, a1, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v28, v8
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: bltu a1, a0, .LBB289_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: .LBB289_2:
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 4
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v0, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v24, v8
; ZVFHMIN-NEXT: vmv8r.v v8, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %b, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %c, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negb, <vscale x 32 x half> %va, <vscale x 32 x half> %negc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vf_nxv32f16(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv32f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv32f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x28, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 40 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 1
; ZVFHMIN-NEXT: mv a3, a0
; ZVFHMIN-NEXT: addi a4, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a4, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a0, a2, .LBB290_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB290_2:
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v8, v0.t
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: mv a4, a3
; ZVFHMIN-NEXT: slli a3, a3, 1
; ZVFHMIN-NEXT: add a3, a3, a4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: sub a2, a0, a2
; ZVFHMIN-NEXT: sltu a0, a0, a2
; ZVFHMIN-NEXT: addi a0, a0, -1
; ZVFHMIN-NEXT: and a0, a0, a2
; ZVFHMIN-NEXT: srli a1, a1, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add a0, a0, a1
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negva = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x i1> %m, i32 %evl)
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negva, <vscale x 32 x half> %vb, <vscale x 32 x half> %negvc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vf_nxv32f16_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv32f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv32f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 1
; ZVFHMIN-NEXT: mv a3, a0
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a4, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: bltu a0, a2, .LBB291_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB291_2:
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v4
; ZVFHMIN-NEXT: sub a2, a0, a2
; ZVFHMIN-NEXT: sltu a0, a0, a2
; ZVFHMIN-NEXT: addi a0, a0, -1
; ZVFHMIN-NEXT: and a0, a0, a2
; ZVFHMIN-NEXT: srli a1, a1, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v8, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negva = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x i1> %m, i32 %evl)
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %negva, <vscale x 32 x half> %negvc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vf_nxv32f16_unmasked(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv32f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv32f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v28
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v28, v8
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: bltu a0, a1, .LBB292_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB292_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v0
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v24, v8
; ZVFHMIN-NEXT: vmv8r.v v8, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negva = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negva, <vscale x 32 x half> %vb, <vscale x 32 x half> %negvc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vf_nxv32f16_unmasked_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv32f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv32f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v24, v8, a1
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v24, v8, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v28
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: vmv1r.v v0, v24
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v28, v8
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: bltu a0, a1, .LBB293_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB293_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v0
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v24, v16
; ZVFHMIN-NEXT: vmv8r.v v8, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negva = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %negva, <vscale x 32 x half> %negvc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vf_nxv32f16_neg_splat(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv32f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv32f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v24, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 1
; ZVFHMIN-NEXT: mv a3, a0
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a4, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: bltu a0, a2, .LBB294_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB294_2:
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: addi a4, sp, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: addi a4, sp, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v28
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v4
; ZVFHMIN-NEXT: sub a2, a0, a2
; ZVFHMIN-NEXT: sltu a0, a0, a2
; ZVFHMIN-NEXT: addi a0, a0, -1
; ZVFHMIN-NEXT: and a0, a0, a2
; ZVFHMIN-NEXT: srli a1, a1, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v8, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %negvb, <vscale x 32 x half> %negvc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vf_nxv32f16_neg_splat_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv32f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv32f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v24, a1, v0.t
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1, v0.t
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 1
; ZVFHMIN-NEXT: mv a3, a0
; ZVFHMIN-NEXT: addi a4, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a4, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a0, a2, .LBB295_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB295_2:
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v28
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v4
; ZVFHMIN-NEXT: sub a2, a0, a2
; ZVFHMIN-NEXT: sltu a0, a0, a2
; ZVFHMIN-NEXT: addi a0, a0, -1
; ZVFHMIN-NEXT: and a0, a0, a2
; ZVFHMIN-NEXT: srli a1, a1, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negvb, <vscale x 32 x half> %va, <vscale x 32 x half> %negvc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vf_nxv32f16_neg_splat_unmasked(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv32f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv32f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v24, v24, a1
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v24, v8, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v28
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: vmv1r.v v0, v24
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v28, v8
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: bltu a0, a1, .LBB296_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB296_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v0
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v24, v16
; ZVFHMIN-NEXT: vmv8r.v v8, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %negvb, <vscale x 32 x half> %negvc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmadd_vf_nxv32f16_neg_splat_unmasked_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmadd_vf_nxv32f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmadd_vf_nxv32f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v24, a1
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v24, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v28, v8
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: bltu a0, a1, .LBB297_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB297_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v0
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v24, v8
; ZVFHMIN-NEXT: vmv8r.v v8, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negvb, <vscale x 32 x half> %va, <vscale x 32 x half> %negvc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv32f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vv v16, v8, v24, v0.t
; ZVFH-NEXT: vmv.v.v v8, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv32f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 5
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: vl8re16.v v24, (a0)
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a0, 8
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v16, a0, v0.t
; ZVFHMIN-NEXT: vxor.vx v24, v24, a0, v0.t
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a2, a0, 1
; ZVFHMIN-NEXT: mv a3, a1
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: vsetvli a4, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a1, a2, .LBB298_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB298_2:
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v4
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: sub a2, a1, a2
; ZVFHMIN-NEXT: sltu a1, a1, a2
; ZVFHMIN-NEXT: addi a1, a1, -1
; ZVFHMIN-NEXT: and a1, a1, a2
; ZVFHMIN-NEXT: srli a0, a0, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v8, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
%negc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %negb, <vscale x 32 x half> %negc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vv_nxv32f16_commuted(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv32f16_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v16, v24, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv32f16_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 2
; ZVFHMIN-NEXT: add a2, a2, a3
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x28, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 40 * vlenb
; ZVFHMIN-NEXT: vl8re16.v v24, (a0)
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a0, 8
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v16, a0, v0.t
; ZVFHMIN-NEXT: vxor.vx v16, v24, a0, v0.t
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a2, a0, 1
; ZVFHMIN-NEXT: mv a3, a1
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: vsetvli a4, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a1, a2, .LBB299_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB299_2:
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v4
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: mv a4, a3
; ZVFHMIN-NEXT: slli a3, a3, 1
; ZVFHMIN-NEXT: add a3, a3, a4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v28
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: sub a2, a1, a2
; ZVFHMIN-NEXT: sltu a1, a1, a2
; ZVFHMIN-NEXT: addi a1, a1, -1
; ZVFHMIN-NEXT: and a1, a1, a2
; ZVFHMIN-NEXT: srli a0, a0, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v8, v16, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v16
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add a0, a0, a1
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
%negc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %c, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negb, <vscale x 32 x half> %va, <vscale x 32 x half> %negc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vv_nxv32f16_unmasked(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv32f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v16, v24
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv32f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 5
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vl8re16.v v24, (a0)
; ZVFHMIN-NEXT: lui a0, 8
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v16, a0
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v24, v24, a0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a0, a2, 1
; ZVFHMIN-NEXT: sub a3, a1, a0
; ZVFHMIN-NEXT: sltu a4, a1, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v28
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v28
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v28, v8
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a1, a0, .LBB300_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: .LBB300_2:
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 4
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v8, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v24, v0
; ZVFHMIN-NEXT: vmv8r.v v8, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %b, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %c, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %negb, <vscale x 32 x half> %negc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vv_nxv32f16_unmasked_commuted(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x half> %c, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vv_nxv32f16_unmasked_commuted:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vl8re16.v v24, (a0)
; ZVFH-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFH-NEXT: vfnmadd.vv v8, v16, v24
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vv_nxv32f16_unmasked_commuted:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 5
; ZVFHMIN-NEXT: sub sp, sp, a2
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vl8re16.v v24, (a0)
; ZVFHMIN-NEXT: lui a0, 8
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v16, a0
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: vsetvli zero, a1, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v24, a0
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a0) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a0, a2, 1
; ZVFHMIN-NEXT: sub a3, a1, a0
; ZVFHMIN-NEXT: sltu a4, a1, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v28, v8
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: bltu a1, a0, .LBB301_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: .LBB301_2:
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a2, a0
; ZVFHMIN-NEXT: slli a0, a0, 1
; ZVFHMIN-NEXT: add a0, a0, a2
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 4
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v0, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v24, v8
; ZVFHMIN-NEXT: vmv8r.v v8, v24
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%negb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %b, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %c, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negb, <vscale x 32 x half> %va, <vscale x 32 x half> %negc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vf_nxv32f16(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv32f16:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv32f16:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 2
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x28, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 40 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 1
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: mv a3, a0
; ZVFHMIN-NEXT: bltu a0, a2, .LBB302_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB302_2:
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v8, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v4
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 4
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: sub a2, a0, a2
; ZVFHMIN-NEXT: sltu a0, a0, a2
; ZVFHMIN-NEXT: addi a0, a0, -1
; ZVFHMIN-NEXT: and a0, a0, a2
; ZVFHMIN-NEXT: srli a1, a1, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 3
; ZVFHMIN-NEXT: mv a1, a0
; ZVFHMIN-NEXT: slli a0, a0, 2
; ZVFHMIN-NEXT: add a0, a0, a1
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negva = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negva, <vscale x 32 x half> %vb, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vf_nxv32f16_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv32f16_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv32f16_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 1
; ZVFHMIN-NEXT: vmv4r.v v4, v12
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
; ZVFHMIN-NEXT: mv a3, a0
; ZVFHMIN-NEXT: bltu a0, a2, .LBB303_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB303_2:
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: sub a2, a0, a2
; ZVFHMIN-NEXT: sltu a0, a0, a2
; ZVFHMIN-NEXT: addi a0, a0, -1
; ZVFHMIN-NEXT: and a0, a0, a2
; ZVFHMIN-NEXT: srli a1, a1, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negva = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %negva, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vf_nxv32f16_unmasked(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv32f16_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv32f16_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a1, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vmv4r.v v8, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v20, v8
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a0, a1, .LBB304_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB304_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v0, v8
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v16, v24
; ZVFHMIN-NEXT: vmv8r.v v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negva = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negva, <vscale x 32 x half> %vb, <vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vf_nxv32f16_unmasked_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv32f16_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv32f16_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a1, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vmv4r.v v8, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v8, v16, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v24
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v16
; ZVFHMIN-NEXT: bltu a0, a1, .LBB305_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB305_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v24, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negva = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %negva, <vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vf_nxv32f16_neg_splat(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv32f16_neg_splat:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv32f16_neg_splat:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vmv8r.v v16, v8
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v8, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs1r.v v0, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1, v0.t
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 1
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
; ZVFHMIN-NEXT: mv a3, a0
; ZVFHMIN-NEXT: bltu a0, a2, .LBB306_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB306_2:
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v8
; ZVFHMIN-NEXT: addi a4, sp, 16
; ZVFHMIN-NEXT: vs8r.v v0, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vmv8r.v v0, v16
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v0
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a4, a4, 4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl1r.v v0, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: addi a4, sp, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v16, v8, v0.t
; ZVFHMIN-NEXT: addi a3, sp, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a3, a3, 3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: sub a2, a0, a2
; ZVFHMIN-NEXT: sltu a0, a0, a2
; ZVFHMIN-NEXT: addi a0, a0, -1
; ZVFHMIN-NEXT: and a0, a0, a2
; ZVFHMIN-NEXT: srli a1, a1, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v4
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v24, v8, v0.t
; ZVFHMIN-NEXT: addi a0, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %negvb, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vf_nxv32f16_neg_splat_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv32f16_neg_splat_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv32f16_neg_splat_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 5
; ZVFHMIN-NEXT: add a1, a2, a1
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x21, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 33 * vlenb
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a2, a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs1r.v v0, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v16, v24, a1, v0.t
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 1
; ZVFHMIN-NEXT: vmv4r.v v28, v20
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a4, a3, 4
; ZVFHMIN-NEXT: add a3, a4, a3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a4, a3, 3
; ZVFHMIN-NEXT: add a3, a4, a3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: mv a3, a0
; ZVFHMIN-NEXT: bltu a0, a2, .LBB307_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a3, a2
; ZVFHMIN-NEXT: .LBB307_2:
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: mv a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 3
; ZVFHMIN-NEXT: add a5, a5, a4
; ZVFHMIN-NEXT: slli a4, a4, 1
; ZVFHMIN-NEXT: add a4, a4, a5
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a4) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vmv8r.v v0, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v0
; ZVFHMIN-NEXT: addi a4, sp, 16
; ZVFHMIN-NEXT: vl1r.v v0, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: slli a5, a4, 3
; ZVFHMIN-NEXT: add a4, a5, a4
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: csrr a4, vlenb
; ZVFHMIN-NEXT: add a4, sp, a4
; ZVFHMIN-NEXT: addi a4, a4, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a4) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a4, a3, 4
; ZVFHMIN-NEXT: add a3, a4, a3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a3) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a3, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: csrr a3, vlenb
; ZVFHMIN-NEXT: slli a4, a3, 3
; ZVFHMIN-NEXT: add a3, a4, a3
; ZVFHMIN-NEXT: add a3, sp, a3
; ZVFHMIN-NEXT: addi a3, a3, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a3) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: sub a2, a0, a2
; ZVFHMIN-NEXT: sltu a0, a0, a2
; ZVFHMIN-NEXT: addi a0, a0, -1
; ZVFHMIN-NEXT: and a0, a0, a2
; ZVFHMIN-NEXT: srli a1, a1, 2
; ZVFHMIN-NEXT: vsetvli a2, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v0, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a2, a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v4
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a2, a1, 3
; ZVFHMIN-NEXT: add a1, a2, a1
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v8, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: add a0, sp, a0
; ZVFHMIN-NEXT: addi a0, a0, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v24
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a1, a0, 5
; ZVFHMIN-NEXT: add a0, a1, a0
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negvb, <vscale x 32 x half> %va, <vscale x 32 x half> %vc, <vscale x 32 x i1> %m, i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vf_nxv32f16_neg_splat_unmasked(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv32f16_neg_splat_unmasked:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv32f16_neg_splat_unmasked:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: vmv8r.v v24, v16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v8, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v8, a1
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a1, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v12
; ZVFHMIN-NEXT: vmv8r.v v8, v24
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v28
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v16, v8, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v12, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v16
; ZVFHMIN-NEXT: bltu a0, a1, .LBB308_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB308_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v16, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v0, v24, v16
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v0
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %negvb, <vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
define <vscale x 32 x half> @vfnmsub_vf_nxv32f16_neg_splat_unmasked_commute(<vscale x 32 x half> %va, half %b, <vscale x 32 x half> %vc, i32 zeroext %evl) {
; ZVFH-LABEL: vfnmsub_vf_nxv32f16_neg_splat_unmasked_commute:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFH-NEXT: vfnmsub.vf v8, fa0, v16
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfnmsub_vf_nxv32f16_neg_splat_unmasked_commute:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: addi sp, sp, -16
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 5
; ZVFHMIN-NEXT: sub sp, sp, a1
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: fmv.x.h a1, fa0
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m8, ta, ma
; ZVFHMIN-NEXT: vmv.v.x v24, a1
; ZVFHMIN-NEXT: lui a1, 8
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
; ZVFHMIN-NEXT: vxor.vx v8, v24, a1
; ZVFHMIN-NEXT: addi a1, sp, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a1) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vsetvli a1, zero, e8, m4, ta, ma
; ZVFHMIN-NEXT: vmset.m v7
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a1, a2, 1
; ZVFHMIN-NEXT: sub a3, a0, a1
; ZVFHMIN-NEXT: sltu a4, a0, a3
; ZVFHMIN-NEXT: addi a4, a4, -1
; ZVFHMIN-NEXT: and a3, a4, a3
; ZVFHMIN-NEXT: srli a2, a2, 2
; ZVFHMIN-NEXT: vsetvli a4, zero, e8, mf2, ta, ma
; ZVFHMIN-NEXT: vslidedown.vx v0, v7, a2
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v12
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v24, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vmv4r.v v8, v16
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: mv a4, a2
; ZVFHMIN-NEXT: slli a2, a2, 1
; ZVFHMIN-NEXT: add a2, a2, a4
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v20
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vl8r.v v16, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a3, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v8, v16, v24, v0.t
; ZVFHMIN-NEXT: vsetvli a2, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v20, v8
; ZVFHMIN-NEXT: addi a2, sp, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a2) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a2, vlenb
; ZVFHMIN-NEXT: slli a2, a2, 3
; ZVFHMIN-NEXT: add a2, sp, a2
; ZVFHMIN-NEXT: addi a2, a2, 16
; ZVFHMIN-NEXT: vs8r.v v8, (a2) # Unknown-size Folded Spill
; ZVFHMIN-NEXT: bltu a0, a1, .LBB309_2
; ZVFHMIN-NEXT: # %bb.1:
; ZVFHMIN-NEXT: mv a0, a1
; ZVFHMIN-NEXT: .LBB309_2:
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 4
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v24, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: mv a2, a1
; ZVFHMIN-NEXT: slli a1, a1, 1
; ZVFHMIN-NEXT: add a1, a1, a2
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v0
; ZVFHMIN-NEXT: csrr a1, vlenb
; ZVFHMIN-NEXT: slli a1, a1, 3
; ZVFHMIN-NEXT: add a1, sp, a1
; ZVFHMIN-NEXT: addi a1, a1, 16
; ZVFHMIN-NEXT: vl8r.v v0, (a1) # Unknown-size Folded Reload
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v24, v0, v8
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, m4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v16, v24
; ZVFHMIN-NEXT: vmv8r.v v8, v16
; ZVFHMIN-NEXT: csrr a0, vlenb
; ZVFHMIN-NEXT: slli a0, a0, 5
; ZVFHMIN-NEXT: add sp, sp, a0
; ZVFHMIN-NEXT: addi sp, sp, 16
; ZVFHMIN-NEXT: ret
%elt.head = insertelement <vscale x 32 x half> poison, half %b, i32 0
%vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
%negvb = call <vscale x 32 x half> @llvm.vp.fneg.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 32 x half> @llvm.vp.fma.nxv32f16(<vscale x 32 x half> %negvb, <vscale x 32 x half> %va, <vscale x 32 x half> %vc, <vscale x 32 x i1> splat (i1 true), i32 %evl)
ret <vscale x 32 x half> %v
}
declare <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float>, <vscale x 1 x i1>, i32)
define <vscale x 1 x float> @vfmsub_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmsub.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv1r.v v8, v9
; CHECK-NEXT: ret
%negc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfmsub_vv_nxv1f32_unmasked(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv1f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmsub.vv v8, v9, v10
; CHECK-NEXT: ret
%negc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfmsub_vf_nxv1f32(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfmsub_vf_nxv1f32_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv1f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %va, <vscale x 1 x float> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfmsub_vf_nxv1f32_unmasked(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv1f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfmsub_vf_nxv1f32_unmasked_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv1f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %va, <vscale x 1 x float> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv1r.v v8, v9
; CHECK-NEXT: ret
%negb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %negb, <vscale x 1 x float> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vv_nxv1f32_commuted(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv1f32_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negb, <vscale x 1 x float> %va, <vscale x 1 x float> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vv_nxv1f32_unmasked(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv1f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %negb, <vscale x 1 x float> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vv_nxv1f32_unmasked_commuted(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv1f32_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negb, <vscale x 1 x float> %va, <vscale x 1 x float> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vf_nxv1f32(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negva, <vscale x 1 x float> %vb, <vscale x 1 x float> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vf_nxv1f32_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %negva, <vscale x 1 x float> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vf_nxv1f32_unmasked(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negva, <vscale x 1 x float> %vb, <vscale x 1 x float> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vf_nxv1f32_unmasked_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %negva, <vscale x 1 x float> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vf_nxv1f32_neg_splat(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f32_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %negvb, <vscale x 1 x float> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vf_nxv1f32_neg_splat_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f32_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negvb, <vscale x 1 x float> %va, <vscale x 1 x float> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vf_nxv1f32_neg_splat_unmasked(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f32_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %negvb, <vscale x 1 x float> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmadd_vf_nxv1f32_neg_splat_unmasked_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f32_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negvb, <vscale x 1 x float> %va, <vscale x 1 x float> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv1r.v v8, v9
; CHECK-NEXT: ret
%negb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %negb, <vscale x 1 x float> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vv_nxv1f32_commuted(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv1f32_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negb, <vscale x 1 x float> %va, <vscale x 1 x float> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vv_nxv1f32_unmasked(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv1f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %negb, <vscale x 1 x float> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vv_nxv1f32_unmasked_commuted(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv1f32_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negb, <vscale x 1 x float> %va, <vscale x 1 x float> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vf_nxv1f32(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negva, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vf_nxv1f32_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %negva, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vf_nxv1f32_unmasked(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negva, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vf_nxv1f32_unmasked_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %negva, <vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vf_nxv1f32_neg_splat(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f32_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %negvb, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vf_nxv1f32_neg_splat_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f32_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negvb, <vscale x 1 x float> %va, <vscale x 1 x float> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vf_nxv1f32_neg_splat_unmasked(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f32_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %negvb, <vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
define <vscale x 1 x float> @vfnmsub_vf_nxv1f32_neg_splat_unmasked_commute(<vscale x 1 x float> %va, float %b, <vscale x 1 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f32_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x float> @llvm.vp.fma.nxv1f32(<vscale x 1 x float> %negvb, <vscale x 1 x float> %va, <vscale x 1 x float> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x float> %v
}
declare <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float>, <vscale x 2 x i1>, i32)
define <vscale x 2 x float> @vfmsub_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmsub.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
%negc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfmsub_vv_nxv2f32_unmasked(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv2f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmsub.vv v8, v9, v10
; CHECK-NEXT: ret
%negc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfmsub_vf_nxv2f32(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfmsub_vf_nxv2f32_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv2f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %va, <vscale x 2 x float> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfmsub_vf_nxv2f32_unmasked(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv2f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfmsub_vf_nxv2f32_unmasked_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv2f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %va, <vscale x 2 x float> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
%negb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %negb, <vscale x 2 x float> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vv_nxv2f32_commuted(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv2f32_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negb, <vscale x 2 x float> %va, <vscale x 2 x float> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vv_nxv2f32_unmasked(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv2f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %negb, <vscale x 2 x float> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vv_nxv2f32_unmasked_commuted(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv2f32_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negb, <vscale x 2 x float> %va, <vscale x 2 x float> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vf_nxv2f32(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negva, <vscale x 2 x float> %vb, <vscale x 2 x float> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vf_nxv2f32_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %negva, <vscale x 2 x float> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vf_nxv2f32_unmasked(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negva, <vscale x 2 x float> %vb, <vscale x 2 x float> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vf_nxv2f32_unmasked_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %negva, <vscale x 2 x float> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vf_nxv2f32_neg_splat(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f32_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %negvb, <vscale x 2 x float> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vf_nxv2f32_neg_splat_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f32_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negvb, <vscale x 2 x float> %va, <vscale x 2 x float> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vf_nxv2f32_neg_splat_unmasked(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f32_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %negvb, <vscale x 2 x float> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmadd_vf_nxv2f32_neg_splat_unmasked_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f32_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negvb, <vscale x 2 x float> %va, <vscale x 2 x float> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
%negb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %negb, <vscale x 2 x float> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vv_nxv2f32_commuted(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv2f32_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negb, <vscale x 2 x float> %va, <vscale x 2 x float> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vv_nxv2f32_unmasked(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv2f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %negb, <vscale x 2 x float> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vv_nxv2f32_unmasked_commuted(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv2f32_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negb, <vscale x 2 x float> %va, <vscale x 2 x float> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vf_nxv2f32(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negva, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vf_nxv2f32_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %negva, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vf_nxv2f32_unmasked(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negva, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vf_nxv2f32_unmasked_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %negva, <vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vf_nxv2f32_neg_splat(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f32_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %negvb, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vf_nxv2f32_neg_splat_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f32_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negvb, <vscale x 2 x float> %va, <vscale x 2 x float> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vf_nxv2f32_neg_splat_unmasked(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f32_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %negvb, <vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
define <vscale x 2 x float> @vfnmsub_vf_nxv2f32_neg_splat_unmasked_commute(<vscale x 2 x float> %va, float %b, <vscale x 2 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f32_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x float> @llvm.vp.fma.nxv2f32(<vscale x 2 x float> %negvb, <vscale x 2 x float> %va, <vscale x 2 x float> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x float> %v
}
declare <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float>, <vscale x 4 x i1>, i32)
define <vscale x 4 x float> @vfmsub_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmsub.vv v10, v8, v12, v0.t
; CHECK-NEXT: vmv.v.v v8, v10
; CHECK-NEXT: ret
%negc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfmsub_vv_nxv4f32_unmasked(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv4f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmsub.vv v8, v10, v12
; CHECK-NEXT: ret
%negc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfmsub_vf_nxv4f32(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfmsub_vf_nxv4f32_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv4f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %va, <vscale x 4 x float> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfmsub_vf_nxv4f32_unmasked(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv4f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfmsub_vf_nxv4f32_unmasked_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv4f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %va, <vscale x 4 x float> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v10, v8, v12, v0.t
; CHECK-NEXT: vmv.v.v v8, v10
; CHECK-NEXT: ret
%negb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %negb, <vscale x 4 x float> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vv_nxv4f32_commuted(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv4f32_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negb, <vscale x 4 x float> %va, <vscale x 4 x float> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vv_nxv4f32_unmasked(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv4f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12
; CHECK-NEXT: ret
%negb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %negb, <vscale x 4 x float> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vv_nxv4f32_unmasked_commuted(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv4f32_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12
; CHECK-NEXT: ret
%negb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negb, <vscale x 4 x float> %va, <vscale x 4 x float> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vf_nxv4f32(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negva, <vscale x 4 x float> %vb, <vscale x 4 x float> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vf_nxv4f32_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %negva, <vscale x 4 x float> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vf_nxv4f32_unmasked(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negva, <vscale x 4 x float> %vb, <vscale x 4 x float> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vf_nxv4f32_unmasked_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %negva, <vscale x 4 x float> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vf_nxv4f32_neg_splat(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f32_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %negvb, <vscale x 4 x float> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vf_nxv4f32_neg_splat_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f32_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negvb, <vscale x 4 x float> %va, <vscale x 4 x float> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vf_nxv4f32_neg_splat_unmasked(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f32_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %negvb, <vscale x 4 x float> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmadd_vf_nxv4f32_neg_splat_unmasked_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f32_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negvb, <vscale x 4 x float> %va, <vscale x 4 x float> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v10, v8, v12, v0.t
; CHECK-NEXT: vmv.v.v v8, v10
; CHECK-NEXT: ret
%negb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %negb, <vscale x 4 x float> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vv_nxv4f32_commuted(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv4f32_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negb, <vscale x 4 x float> %va, <vscale x 4 x float> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vv_nxv4f32_unmasked(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv4f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12
; CHECK-NEXT: ret
%negb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %negb, <vscale x 4 x float> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vv_nxv4f32_unmasked_commuted(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv4f32_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12
; CHECK-NEXT: ret
%negb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negb, <vscale x 4 x float> %va, <vscale x 4 x float> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vf_nxv4f32(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negva, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vf_nxv4f32_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %negva, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vf_nxv4f32_unmasked(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negva, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vf_nxv4f32_unmasked_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %negva, <vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vf_nxv4f32_neg_splat(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f32_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %negvb, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vf_nxv4f32_neg_splat_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f32_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negvb, <vscale x 4 x float> %va, <vscale x 4 x float> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vf_nxv4f32_neg_splat_unmasked(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f32_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %negvb, <vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
define <vscale x 4 x float> @vfnmsub_vf_nxv4f32_neg_splat_unmasked_commute(<vscale x 4 x float> %va, float %b, <vscale x 4 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f32_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x float> @llvm.vp.fma.nxv4f32(<vscale x 4 x float> %negvb, <vscale x 4 x float> %va, <vscale x 4 x float> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x float> %v
}
declare <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float>, <vscale x 8 x i1>, i32)
define <vscale x 8 x float> @vfmsub_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv8f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmsub.vv v12, v8, v16, v0.t
; CHECK-NEXT: vmv.v.v v8, v12
; CHECK-NEXT: ret
%negc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfmsub_vv_nxv8f32_unmasked(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv8f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmsub.vv v8, v12, v16
; CHECK-NEXT: ret
%negc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfmsub_vf_nxv8f32(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv8f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfmsub_vf_nxv8f32_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv8f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %va, <vscale x 8 x float> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfmsub_vf_nxv8f32_unmasked(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv8f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfmsub_vf_nxv8f32_unmasked_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv8f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %va, <vscale x 8 x float> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv8f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v12, v8, v16, v0.t
; CHECK-NEXT: vmv.v.v v8, v12
; CHECK-NEXT: ret
%negb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %negb, <vscale x 8 x float> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vv_nxv8f32_commuted(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv8f32_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negb, <vscale x 8 x float> %va, <vscale x 8 x float> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vv_nxv8f32_unmasked(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv8f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16
; CHECK-NEXT: ret
%negb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %negb, <vscale x 8 x float> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vv_nxv8f32_unmasked_commuted(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv8f32_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16
; CHECK-NEXT: ret
%negb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negb, <vscale x 8 x float> %va, <vscale x 8 x float> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vf_nxv8f32(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negva, <vscale x 8 x float> %vb, <vscale x 8 x float> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vf_nxv8f32_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %negva, <vscale x 8 x float> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vf_nxv8f32_unmasked(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negva, <vscale x 8 x float> %vb, <vscale x 8 x float> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vf_nxv8f32_unmasked_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %negva, <vscale x 8 x float> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vf_nxv8f32_neg_splat(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f32_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %negvb, <vscale x 8 x float> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vf_nxv8f32_neg_splat_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f32_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negvb, <vscale x 8 x float> %va, <vscale x 8 x float> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vf_nxv8f32_neg_splat_unmasked(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f32_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %negvb, <vscale x 8 x float> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmadd_vf_nxv8f32_neg_splat_unmasked_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f32_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negvb, <vscale x 8 x float> %va, <vscale x 8 x float> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv8f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v12, v8, v16, v0.t
; CHECK-NEXT: vmv.v.v v8, v12
; CHECK-NEXT: ret
%negb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %negb, <vscale x 8 x float> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vv_nxv8f32_commuted(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv8f32_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negb, <vscale x 8 x float> %va, <vscale x 8 x float> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vv_nxv8f32_unmasked(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv8f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16
; CHECK-NEXT: ret
%negb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %negb, <vscale x 8 x float> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vv_nxv8f32_unmasked_commuted(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv8f32_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16
; CHECK-NEXT: ret
%negb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negb, <vscale x 8 x float> %va, <vscale x 8 x float> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vf_nxv8f32(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negva, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vf_nxv8f32_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %negva, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vf_nxv8f32_unmasked(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negva, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vf_nxv8f32_unmasked_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %negva, <vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vf_nxv8f32_neg_splat(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f32_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %negvb, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vf_nxv8f32_neg_splat_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f32_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negvb, <vscale x 8 x float> %va, <vscale x 8 x float> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vf_nxv8f32_neg_splat_unmasked(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f32_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %negvb, <vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
define <vscale x 8 x float> @vfnmsub_vf_nxv8f32_neg_splat_unmasked_commute(<vscale x 8 x float> %va, float %b, <vscale x 8 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f32_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x float> @llvm.vp.fma.nxv8f32(<vscale x 8 x float> %negvb, <vscale x 8 x float> %va, <vscale x 8 x float> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x float> %v
}
declare <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float>, <vscale x 16 x i1>, i32)
define <vscale x 16 x float> @vfmsub_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv16f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfmsub.vv v16, v8, v24, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: ret
%negc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %negc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfmsub_vv_nxv16f32_unmasked(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv16f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfmsub.vv v8, v16, v24
; CHECK-NEXT: ret
%negc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %negc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfmsub_vf_nxv16f32(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv16f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x float> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfmsub_vf_nxv16f32_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv16f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %va, <vscale x 16 x float> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfmsub_vf_nxv16f32_unmasked(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv16f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x float> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfmsub_vf_nxv16f32_unmasked_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv16f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %va, <vscale x 16 x float> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv16f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v16, v8, v24, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: ret
%negb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
%negc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %negb, <vscale x 16 x float> %negc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vv_nxv16f32_commuted(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv16f32_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
%negc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negb, <vscale x 16 x float> %va, <vscale x 16 x float> %negc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vv_nxv16f32_unmasked(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv16f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24
; CHECK-NEXT: ret
%negb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %b, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %negb, <vscale x 16 x float> %negc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vv_nxv16f32_unmasked_commuted(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv16f32_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24
; CHECK-NEXT: ret
%negb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %b, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negb, <vscale x 16 x float> %va, <vscale x 16 x float> %negc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vf_nxv16f32(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv16f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x i1> %m, i32 %evl)
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negva, <vscale x 16 x float> %vb, <vscale x 16 x float> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vf_nxv16f32_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv16f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x i1> %m, i32 %evl)
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %negva, <vscale x 16 x float> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vf_nxv16f32_unmasked(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv16f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negva, <vscale x 16 x float> %vb, <vscale x 16 x float> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vf_nxv16f32_unmasked_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv16f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %negva, <vscale x 16 x float> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vf_nxv16f32_neg_splat(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv16f32_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %negvb, <vscale x 16 x float> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vf_nxv16f32_neg_splat_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv16f32_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negvb, <vscale x 16 x float> %va, <vscale x 16 x float> %negvc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vf_nxv16f32_neg_splat_unmasked(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv16f32_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %negvb, <vscale x 16 x float> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmadd_vf_nxv16f32_neg_splat_unmasked_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv16f32_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negvb, <vscale x 16 x float> %va, <vscale x 16 x float> %negvc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv16f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v16, v8, v24, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: ret
%negb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
%negc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %negb, <vscale x 16 x float> %negc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vv_nxv16f32_commuted(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv16f32_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
%negc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %c, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negb, <vscale x 16 x float> %va, <vscale x 16 x float> %negc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vv_nxv16f32_unmasked(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv16f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24
; CHECK-NEXT: ret
%negb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %b, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %negb, <vscale x 16 x float> %negc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vv_nxv16f32_unmasked_commuted(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x float> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv16f32_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re32.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e32, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24
; CHECK-NEXT: ret
%negb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %b, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %c, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negb, <vscale x 16 x float> %va, <vscale x 16 x float> %negc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vf_nxv16f32(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv16f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negva, <vscale x 16 x float> %vb, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vf_nxv16f32_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv16f32_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %negva, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vf_nxv16f32_unmasked(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv16f32_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negva, <vscale x 16 x float> %vb, <vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vf_nxv16f32_unmasked_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv16f32_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negva = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %negva, <vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vf_nxv16f32_neg_splat(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv16f32_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %negvb, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vf_nxv16f32_neg_splat_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv16f32_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negvb, <vscale x 16 x float> %va, <vscale x 16 x float> %vc, <vscale x 16 x i1> %m, i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vf_nxv16f32_neg_splat_unmasked(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv16f32_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %negvb, <vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
define <vscale x 16 x float> @vfnmsub_vf_nxv16f32_neg_splat_unmasked_commute(<vscale x 16 x float> %va, float %b, <vscale x 16 x float> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv16f32_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e32, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 16 x float> poison, float %b, i32 0
%vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
%negvb = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 16 x float> @llvm.vp.fma.nxv16f32(<vscale x 16 x float> %negvb, <vscale x 16 x float> %va, <vscale x 16 x float> %vc, <vscale x 16 x i1> splat (i1 true), i32 %evl)
ret <vscale x 16 x float> %v
}
declare <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double>, <vscale x 1 x i1>, i32)
define <vscale x 1 x double> @vfmsub_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmsub.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
%negc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfmsub_vv_nxv1f64_unmasked(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv1f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmsub.vv v8, v9, v10
; CHECK-NEXT: ret
%negc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfmsub_vf_nxv1f64(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x double> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfmsub_vf_nxv1f64_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv1f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %va, <vscale x 1 x double> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfmsub_vf_nxv1f64_unmasked(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv1f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x double> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfmsub_vf_nxv1f64_unmasked_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv1f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %va, <vscale x 1 x double> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
%negb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %negb, <vscale x 1 x double> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vv_nxv1f64_commuted(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv1f64_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negb, <vscale x 1 x double> %va, <vscale x 1 x double> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vv_nxv1f64_unmasked(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv1f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %negb, <vscale x 1 x double> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vv_nxv1f64_unmasked_commuted(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv1f64_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negb, <vscale x 1 x double> %va, <vscale x 1 x double> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vf_nxv1f64(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negva, <vscale x 1 x double> %vb, <vscale x 1 x double> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vf_nxv1f64_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %negva, <vscale x 1 x double> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vf_nxv1f64_unmasked(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negva, <vscale x 1 x double> %vb, <vscale x 1 x double> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vf_nxv1f64_unmasked_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %negva, <vscale x 1 x double> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vf_nxv1f64_neg_splat(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f64_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %negvb, <vscale x 1 x double> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vf_nxv1f64_neg_splat_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f64_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negvb, <vscale x 1 x double> %va, <vscale x 1 x double> %negvc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vf_nxv1f64_neg_splat_unmasked(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f64_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %negvb, <vscale x 1 x double> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmadd_vf_nxv1f64_neg_splat_unmasked_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv1f64_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negvb, <vscale x 1 x double> %va, <vscale x 1 x double> %negvc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v9, v8, v10, v0.t
; CHECK-NEXT: vmv.v.v v8, v9
; CHECK-NEXT: ret
%negb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %negb, <vscale x 1 x double> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vv_nxv1f64_commuted(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv1f64_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
%negc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negb, <vscale x 1 x double> %va, <vscale x 1 x double> %negc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vv_nxv1f64_unmasked(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv1f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %negb, <vscale x 1 x double> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vv_nxv1f64_unmasked_commuted(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv1f64_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v9, v10
; CHECK-NEXT: ret
%negb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %b, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %c, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negb, <vscale x 1 x double> %va, <vscale x 1 x double> %negc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vf_nxv1f64(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negva, <vscale x 1 x double> %vb, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vf_nxv1f64_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %negva, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vf_nxv1f64_unmasked(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negva, <vscale x 1 x double> %vb, <vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vf_nxv1f64_unmasked_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negva = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %negva, <vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vf_nxv1f64_neg_splat(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f64_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %negvb, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vf_nxv1f64_neg_splat_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f64_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negvb, <vscale x 1 x double> %va, <vscale x 1 x double> %vc, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vf_nxv1f64_neg_splat_unmasked(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f64_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %negvb, <vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
define <vscale x 1 x double> @vfnmsub_vf_nxv1f64_neg_splat_unmasked_commute(<vscale x 1 x double> %va, double %b, <vscale x 1 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv1f64_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v9
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 1 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
%negvb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %negvb, <vscale x 1 x double> %va, <vscale x 1 x double> %vc, <vscale x 1 x i1> splat (i1 true), i32 %evl)
ret <vscale x 1 x double> %v
}
declare <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double>, <vscale x 2 x i1>, i32)
define <vscale x 2 x double> @vfmsub_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmsub.vv v10, v8, v12, v0.t
; CHECK-NEXT: vmv.v.v v8, v10
; CHECK-NEXT: ret
%negc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfmsub_vv_nxv2f64_unmasked(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv2f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmsub.vv v8, v10, v12
; CHECK-NEXT: ret
%negc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfmsub_vf_nxv2f64(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x double> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfmsub_vf_nxv2f64_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv2f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %va, <vscale x 2 x double> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfmsub_vf_nxv2f64_unmasked(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv2f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x double> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfmsub_vf_nxv2f64_unmasked_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv2f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %va, <vscale x 2 x double> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v10, v8, v12, v0.t
; CHECK-NEXT: vmv.v.v v8, v10
; CHECK-NEXT: ret
%negb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %negb, <vscale x 2 x double> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vv_nxv2f64_commuted(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv2f64_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negb, <vscale x 2 x double> %va, <vscale x 2 x double> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vv_nxv2f64_unmasked(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv2f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12
; CHECK-NEXT: ret
%negb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %negb, <vscale x 2 x double> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vv_nxv2f64_unmasked_commuted(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv2f64_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12
; CHECK-NEXT: ret
%negb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negb, <vscale x 2 x double> %va, <vscale x 2 x double> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vf_nxv2f64(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negva, <vscale x 2 x double> %vb, <vscale x 2 x double> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vf_nxv2f64_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %negva, <vscale x 2 x double> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vf_nxv2f64_unmasked(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negva, <vscale x 2 x double> %vb, <vscale x 2 x double> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vf_nxv2f64_unmasked_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %negva, <vscale x 2 x double> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vf_nxv2f64_neg_splat(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f64_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %negvb, <vscale x 2 x double> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vf_nxv2f64_neg_splat_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f64_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negvb, <vscale x 2 x double> %va, <vscale x 2 x double> %negvc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vf_nxv2f64_neg_splat_unmasked(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f64_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %negvb, <vscale x 2 x double> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmadd_vf_nxv2f64_neg_splat_unmasked_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv2f64_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negvb, <vscale x 2 x double> %va, <vscale x 2 x double> %negvc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v10, v8, v12, v0.t
; CHECK-NEXT: vmv.v.v v8, v10
; CHECK-NEXT: ret
%negb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %negb, <vscale x 2 x double> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vv_nxv2f64_commuted(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv2f64_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
%negc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %c, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negb, <vscale x 2 x double> %va, <vscale x 2 x double> %negc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vv_nxv2f64_unmasked(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv2f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12
; CHECK-NEXT: ret
%negb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %negb, <vscale x 2 x double> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vv_nxv2f64_unmasked_commuted(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv2f64_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v10, v12
; CHECK-NEXT: ret
%negb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %b, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %c, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negb, <vscale x 2 x double> %va, <vscale x 2 x double> %negc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vf_nxv2f64(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negva, <vscale x 2 x double> %vb, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vf_nxv2f64_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %negva, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vf_nxv2f64_unmasked(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negva, <vscale x 2 x double> %vb, <vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vf_nxv2f64_unmasked_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negva = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %negva, <vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vf_nxv2f64_neg_splat(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f64_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %negvb, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vf_nxv2f64_neg_splat_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f64_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negvb, <vscale x 2 x double> %va, <vscale x 2 x double> %vc, <vscale x 2 x i1> %m, i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vf_nxv2f64_neg_splat_unmasked(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f64_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %negvb, <vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
define <vscale x 2 x double> @vfnmsub_vf_nxv2f64_neg_splat_unmasked_commute(<vscale x 2 x double> %va, double %b, <vscale x 2 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv2f64_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m2, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v10
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 2 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
%negvb = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 2 x double> @llvm.vp.fma.nxv2f64(<vscale x 2 x double> %negvb, <vscale x 2 x double> %va, <vscale x 2 x double> %vc, <vscale x 2 x i1> splat (i1 true), i32 %evl)
ret <vscale x 2 x double> %v
}
declare <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double>, <vscale x 4 x i1>, i32)
define <vscale x 4 x double> @vfmsub_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmsub.vv v12, v8, v16, v0.t
; CHECK-NEXT: vmv.v.v v8, v12
; CHECK-NEXT: ret
%negc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfmsub_vv_nxv4f64_unmasked(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv4f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmsub.vv v8, v12, v16
; CHECK-NEXT: ret
%negc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfmsub_vf_nxv4f64(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x double> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfmsub_vf_nxv4f64_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv4f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %va, <vscale x 4 x double> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfmsub_vf_nxv4f64_unmasked(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv4f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x double> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfmsub_vf_nxv4f64_unmasked_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv4f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %va, <vscale x 4 x double> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v12, v8, v16, v0.t
; CHECK-NEXT: vmv.v.v v8, v12
; CHECK-NEXT: ret
%negb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %negb, <vscale x 4 x double> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vv_nxv4f64_commuted(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv4f64_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negb, <vscale x 4 x double> %va, <vscale x 4 x double> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vv_nxv4f64_unmasked(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv4f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16
; CHECK-NEXT: ret
%negb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %negb, <vscale x 4 x double> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vv_nxv4f64_unmasked_commuted(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv4f64_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16
; CHECK-NEXT: ret
%negb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negb, <vscale x 4 x double> %va, <vscale x 4 x double> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vf_nxv4f64(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negva, <vscale x 4 x double> %vb, <vscale x 4 x double> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vf_nxv4f64_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %negva, <vscale x 4 x double> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vf_nxv4f64_unmasked(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negva, <vscale x 4 x double> %vb, <vscale x 4 x double> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vf_nxv4f64_unmasked_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %negva, <vscale x 4 x double> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vf_nxv4f64_neg_splat(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f64_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %negvb, <vscale x 4 x double> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vf_nxv4f64_neg_splat_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f64_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negvb, <vscale x 4 x double> %va, <vscale x 4 x double> %negvc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vf_nxv4f64_neg_splat_unmasked(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f64_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %negvb, <vscale x 4 x double> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmadd_vf_nxv4f64_neg_splat_unmasked_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv4f64_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negvb, <vscale x 4 x double> %va, <vscale x 4 x double> %negvc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v12, v8, v16, v0.t
; CHECK-NEXT: vmv.v.v v8, v12
; CHECK-NEXT: ret
%negb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %negb, <vscale x 4 x double> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vv_nxv4f64_commuted(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv4f64_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
%negc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %c, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negb, <vscale x 4 x double> %va, <vscale x 4 x double> %negc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vv_nxv4f64_unmasked(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv4f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16
; CHECK-NEXT: ret
%negb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %negb, <vscale x 4 x double> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vv_nxv4f64_unmasked_commuted(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv4f64_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v12, v16
; CHECK-NEXT: ret
%negb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %b, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %c, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negb, <vscale x 4 x double> %va, <vscale x 4 x double> %negc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vf_nxv4f64(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negva, <vscale x 4 x double> %vb, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vf_nxv4f64_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %negva, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vf_nxv4f64_unmasked(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negva, <vscale x 4 x double> %vb, <vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vf_nxv4f64_unmasked_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negva = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %negva, <vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vf_nxv4f64_neg_splat(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f64_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %negvb, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vf_nxv4f64_neg_splat_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f64_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negvb, <vscale x 4 x double> %va, <vscale x 4 x double> %vc, <vscale x 4 x i1> %m, i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vf_nxv4f64_neg_splat_unmasked(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f64_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %negvb, <vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
define <vscale x 4 x double> @vfnmsub_vf_nxv4f64_neg_splat_unmasked_commute(<vscale x 4 x double> %va, double %b, <vscale x 4 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv4f64_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m4, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v12
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 4 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
%negvb = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 4 x double> @llvm.vp.fma.nxv4f64(<vscale x 4 x double> %negvb, <vscale x 4 x double> %va, <vscale x 4 x double> %vc, <vscale x 4 x i1> splat (i1 true), i32 %evl)
ret <vscale x 4 x double> %v
}
declare <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double>, <vscale x 8 x i1>, i32)
define <vscale x 8 x double> @vfmsub_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv8f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfmsub.vv v16, v8, v24, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: ret
%negc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfmsub_vv_nxv8f64_unmasked(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vv_nxv8f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfmsub.vv v8, v16, v24
; CHECK-NEXT: ret
%negc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfmsub_vf_nxv8f64(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv8f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x double> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfmsub_vf_nxv8f64_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv8f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %va, <vscale x 8 x double> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfmsub_vf_nxv8f64_unmasked(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv8f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x double> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfmsub_vf_nxv8f64_unmasked_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfmsub_vf_nxv8f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %va, <vscale x 8 x double> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv8f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v16, v8, v24, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: ret
%negb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %negb, <vscale x 8 x double> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vv_nxv8f64_commuted(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv8f64_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negb, <vscale x 8 x double> %va, <vscale x 8 x double> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vv_nxv8f64_unmasked(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv8f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24
; CHECK-NEXT: ret
%negb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %negb, <vscale x 8 x double> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vv_nxv8f64_unmasked_commuted(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vv_nxv8f64_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24
; CHECK-NEXT: ret
%negb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negb, <vscale x 8 x double> %va, <vscale x 8 x double> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vf_nxv8f64(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negva, <vscale x 8 x double> %vb, <vscale x 8 x double> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vf_nxv8f64_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %negva, <vscale x 8 x double> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vf_nxv8f64_unmasked(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negva, <vscale x 8 x double> %vb, <vscale x 8 x double> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vf_nxv8f64_unmasked_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %negva, <vscale x 8 x double> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vf_nxv8f64_neg_splat(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f64_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %negvb, <vscale x 8 x double> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vf_nxv8f64_neg_splat_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f64_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negvb, <vscale x 8 x double> %va, <vscale x 8 x double> %negvc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vf_nxv8f64_neg_splat_unmasked(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f64_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %negvb, <vscale x 8 x double> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmadd_vf_nxv8f64_neg_splat_unmasked_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmadd_vf_nxv8f64_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negvc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negvb, <vscale x 8 x double> %va, <vscale x 8 x double> %negvc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv8f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v16, v8, v24, v0.t
; CHECK-NEXT: vmv.v.v v8, v16
; CHECK-NEXT: ret
%negb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %negb, <vscale x 8 x double> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vv_nxv8f64_commuted(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv8f64_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24, v0.t
; CHECK-NEXT: ret
%negb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
%negc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %c, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negb, <vscale x 8 x double> %va, <vscale x 8 x double> %negc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vv_nxv8f64_unmasked(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv8f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24
; CHECK-NEXT: ret
%negb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %negb, <vscale x 8 x double> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vv_nxv8f64_unmasked_commuted(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x double> %c, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vv_nxv8f64_unmasked_commuted:
; CHECK: # %bb.0:
; CHECK-NEXT: vl8re64.v v24, (a0)
; CHECK-NEXT: vsetvli zero, a1, e64, m8, ta, ma
; CHECK-NEXT: vfnmadd.vv v8, v16, v24
; CHECK-NEXT: ret
%negb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %b, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%negc = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %c, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negb, <vscale x 8 x double> %va, <vscale x 8 x double> %negc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vf_nxv8f64(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negva, <vscale x 8 x double> %vb, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vf_nxv8f64_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f64_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %negva, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vf_nxv8f64_unmasked(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f64_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negva, <vscale x 8 x double> %vb, <vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vf_nxv8f64_unmasked_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f64_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negva = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %negva, <vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vf_nxv8f64_neg_splat(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f64_neg_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %negvb, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vf_nxv8f64_neg_splat_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f64_neg_splat_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16, v0.t
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negvb, <vscale x 8 x double> %va, <vscale x 8 x double> %vc, <vscale x 8 x i1> %m, i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vf_nxv8f64_neg_splat_unmasked(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f64_neg_splat_unmasked:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %negvb, <vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 8 x double> @vfnmsub_vf_nxv8f64_neg_splat_unmasked_commute(<vscale x 8 x double> %va, double %b, <vscale x 8 x double> %vc, i32 zeroext %evl) {
; CHECK-LABEL: vfnmsub_vf_nxv8f64_neg_splat_unmasked_commute:
; CHECK: # %bb.0:
; CHECK-NEXT: vsetvli zero, a0, e64, m8, ta, ma
; CHECK-NEXT: vfnmsub.vf v8, fa0, v16
; CHECK-NEXT: ret
%elt.head = insertelement <vscale x 8 x double> poison, double %b, i32 0
%vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
%negvb = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x i1> splat (i1 true), i32 %evl)
%v = call <vscale x 8 x double> @llvm.vp.fma.nxv8f64(<vscale x 8 x double> %negvb, <vscale x 8 x double> %va, <vscale x 8 x double> %vc, <vscale x 8 x i1> splat (i1 true), i32 %evl)
ret <vscale x 8 x double> %v
}
define <vscale x 1 x half> @vfma_vv_nxv1f16_double_neg(<vscale x 1 x half> %a, <vscale x 1 x half> %b, <vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; ZVFH-LABEL: vfma_vv_nxv1f16_double_neg:
; ZVFH: # %bb.0:
; ZVFH-NEXT: vsetvli zero, a0, e16, mf4, ta, ma
; ZVFH-NEXT: vfmadd.vv v9, v8, v10, v0.t
; ZVFH-NEXT: vmv1r.v v8, v9
; ZVFH-NEXT: ret
;
; ZVFHMIN-LABEL: vfma_vv_nxv1f16_double_neg:
; ZVFHMIN: # %bb.0:
; ZVFHMIN-NEXT: vsetvli a1, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfwcvt.f.f.v v11, v10
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v9
; ZVFHMIN-NEXT: vsetvli zero, a0, e32, mf2, ta, ma
; ZVFHMIN-NEXT: vfmadd.vv v12, v10, v11, v0.t
; ZVFHMIN-NEXT: vsetvli a0, zero, e16, mf4, ta, ma
; ZVFHMIN-NEXT: vfncvt.f.f.w v8, v12
; ZVFHMIN-NEXT: ret
%nega = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %a, <vscale x 1 x i1> %m, i32 %evl)
%negb = call <vscale x 1 x half> @llvm.vp.fneg.nxv1f16(<vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
%v = call <vscale x 1 x half> @llvm.vp.fma.nxv1f16(<vscale x 1 x half> %nega, <vscale x 1 x half> %negb, <vscale x 1 x half> %c, <vscale x 1 x i1> %m, i32 %evl)
ret <vscale x 1 x half> %v
}