Codebase Browser
llvm
Go to App

llvm/llvm/test/CodeGen/X86/vector-lrint.ll 

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=i686-unknown -mattr=sse2 | FileCheck %s --check-prefix=X86-SSE2
; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=i686-unknown -mattr=avx | FileCheck %s --check-prefixes=X86-AVX,X86-AVX1
; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=i686-unknown -mattr=avx512f | FileCheck %s --check-prefixes=X86-AVX,AVX512-i32
; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=i686-unknown -mattr=avx512dq,avx512vl | FileCheck %s --check-prefixes=X86-AVX,AVX512-i32
; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=x86_64-unknown -mattr=avx | FileCheck %s --check-prefixes=X64-AVX-i32,X64-AVX1-i32
; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=x86_64-unknown -mattr=avx512f | FileCheck %s --check-prefixes=X64-AVX-i32,AVX512-i32
; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=x86_64-unknown -mattr=avx512dq,avx512vl | FileCheck %s --check-prefixes=X64-AVX-i32,AVX512-i32
; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=x86_64-unknown -mattr=avx | FileCheck %s --check-prefixes=X64-AVX-i64,X64-AVX1-i64
; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=x86_64-unknown -mattr=avx512f | FileCheck %s --check-prefixes=X64-AVX-i64,AVX512-i64
; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=x86_64-unknown -mattr=avx512dq,avx512vl | FileCheck %s --check-prefixes=X64-AVX-i64,AVX512DQ-i64

define <1 x iXLen> @lrint_v1f32(<1 x float> %x) {
; X86-SSE2-LABEL: lrint_v1f32:
; X86-SSE2:       # %bb.0:
; X86-SSE2-NEXT:    cvtss2si {{[0-9]+}}(%esp), %eax
; X86-SSE2-NEXT:    retl
;
; X86-AVX-LABEL: lrint_v1f32:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    vcvtss2si {{[0-9]+}}(%esp), %eax
; X86-AVX-NEXT:    retl
;
; X64-AVX-i32-LABEL: lrint_v1f32:
; X64-AVX-i32:       # %bb.0:
; X64-AVX-i32-NEXT:    vcvtss2si %xmm0, %eax
; X64-AVX-i32-NEXT:    retq
;
; X64-AVX-i64-LABEL: lrint_v1f32:
; X64-AVX-i64:       # %bb.0:
; X64-AVX-i64-NEXT:    vcvtss2si %xmm0, %rax
; X64-AVX-i64-NEXT:    retq
  %a = call <1 x iXLen> @llvm.lrint.v1iXLen.v1f32(<1 x float> %x)
  ret <1 x iXLen> %a
}
declare <1 x iXLen> @llvm.lrint.v1iXLen.v1f32(<1 x float>)

define <2 x iXLen> @lrint_v2f32(<2 x float> %x) {
; X86-SSE2-LABEL: lrint_v2f32:
; X86-SSE2:       # %bb.0:
; X86-SSE2-NEXT:    cvtps2dq %xmm0, %xmm0
; X86-SSE2-NEXT:    retl
;
; X86-AVX-LABEL: lrint_v2f32:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    vcvtps2dq %xmm0, %xmm0
; X86-AVX-NEXT:    retl
;
; X64-AVX-i32-LABEL: lrint_v2f32:
; X64-AVX-i32:       # %bb.0:
; X64-AVX-i32-NEXT:    vcvtps2dq %xmm0, %xmm0
; X64-AVX-i32-NEXT:    retq
;
; X64-AVX1-i64-LABEL: lrint_v2f32:
; X64-AVX1-i64:       # %bb.0:
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm1
; X64-AVX1-i64-NEXT:    vmovshdup {{.*#+}} xmm0 = xmm0[1,1,3,3]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm0
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; X64-AVX1-i64-NEXT:    retq
;
; AVX512-i64-LABEL: lrint_v2f32:
; AVX512-i64:       # %bb.0:
; AVX512-i64-NEXT:    vcvtss2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm1
; AVX512-i64-NEXT:    vmovshdup {{.*#+}} xmm0 = xmm0[1,1,3,3]
; AVX512-i64-NEXT:    vcvtss2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm0
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX512-i64-NEXT:    retq
;
; AVX512DQ-i64-LABEL: lrint_v2f32:
; AVX512DQ-i64:       # %bb.0:
; AVX512DQ-i64-NEXT:    vcvtps2qq %xmm0, %xmm0
; AVX512DQ-i64-NEXT:    retq
  %a = call <2 x iXLen> @llvm.lrint.v2iXLen.v2f32(<2 x float> %x)
  ret <2 x iXLen> %a
}
declare <2 x iXLen> @llvm.lrint.v2iXLen.v2f32(<2 x float>)

define <4 x iXLen> @lrint_v4f32(<4 x float> %x) {
; X86-SSE2-LABEL: lrint_v4f32:
; X86-SSE2:       # %bb.0:
; X86-SSE2-NEXT:    cvtps2dq %xmm0, %xmm0
; X86-SSE2-NEXT:    retl
;
; X86-AVX-LABEL: lrint_v4f32:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    vcvtps2dq %xmm0, %xmm0
; X86-AVX-NEXT:    retl
;
; X64-AVX-i32-LABEL: lrint_v4f32:
; X64-AVX-i32:       # %bb.0:
; X64-AVX-i32-NEXT:    vcvtps2dq %xmm0, %xmm0
; X64-AVX-i32-NEXT:    retq
;
; X64-AVX1-i64-LABEL: lrint_v4f32:
; X64-AVX1-i64:       # %bb.0:
; X64-AVX1-i64-NEXT:    vshufps {{.*#+}} xmm1 = xmm0[3,3,3,3]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm1, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm1
; X64-AVX1-i64-NEXT:    vshufpd {{.*#+}} xmm2 = xmm0[1,0]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm2, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm2
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm2
; X64-AVX1-i64-NEXT:    vmovshdup {{.*#+}} xmm0 = xmm0[1,1,3,3]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm0
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm2[0],xmm0[0]
; X64-AVX1-i64-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; X64-AVX1-i64-NEXT:    retq
;
; AVX512-i64-LABEL: lrint_v4f32:
; AVX512-i64:       # %bb.0:
; AVX512-i64-NEXT:    vshufps {{.*#+}} xmm1 = xmm0[3,3,3,3]
; AVX512-i64-NEXT:    vcvtss2si %xmm1, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm1
; AVX512-i64-NEXT:    vshufpd {{.*#+}} xmm2 = xmm0[1,0]
; AVX512-i64-NEXT:    vcvtss2si %xmm2, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm2
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
; AVX512-i64-NEXT:    vcvtss2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm2
; AVX512-i64-NEXT:    vmovshdup {{.*#+}} xmm0 = xmm0[1,1,3,3]
; AVX512-i64-NEXT:    vcvtss2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm0
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm2[0],xmm0[0]
; AVX512-i64-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX512-i64-NEXT:    retq
;
; AVX512DQ-i64-LABEL: lrint_v4f32:
; AVX512DQ-i64:       # %bb.0:
; AVX512DQ-i64-NEXT:    vcvtps2qq %xmm0, %ymm0
; AVX512DQ-i64-NEXT:    retq
  %a = call <4 x iXLen> @llvm.lrint.v4iXLen.v4f32(<4 x float> %x)
  ret <4 x iXLen> %a
}
declare <4 x iXLen> @llvm.lrint.v4iXLen.v4f32(<4 x float>)

define <8 x iXLen> @lrint_v8f32(<8 x float> %x) {
; X86-SSE2-LABEL: lrint_v8f32:
; X86-SSE2:       # %bb.0:
; X86-SSE2-NEXT:    cvtps2dq %xmm0, %xmm0
; X86-SSE2-NEXT:    cvtps2dq %xmm1, %xmm1
; X86-SSE2-NEXT:    retl
;
; X86-AVX-LABEL: lrint_v8f32:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    vcvtps2dq %ymm0, %ymm0
; X86-AVX-NEXT:    retl
;
; X64-AVX-i32-LABEL: lrint_v8f32:
; X64-AVX-i32:       # %bb.0:
; X64-AVX-i32-NEXT:    vcvtps2dq %ymm0, %ymm0
; X64-AVX-i32-NEXT:    retq
;
; X64-AVX1-i64-LABEL: lrint_v8f32:
; X64-AVX1-i64:       # %bb.0:
; X64-AVX1-i64-NEXT:    vshufps {{.*#+}} xmm1 = xmm0[3,3,3,3]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm1, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm1
; X64-AVX1-i64-NEXT:    vshufpd {{.*#+}} xmm2 = xmm0[1,0]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm2, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm2
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm2
; X64-AVX1-i64-NEXT:    vmovshdup {{.*#+}} xmm3 = xmm0[1,1,3,3]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm3, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm3
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm3[0]
; X64-AVX1-i64-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm2
; X64-AVX1-i64-NEXT:    vextractf128 $1, %ymm0, %xmm0
; X64-AVX1-i64-NEXT:    vshufps {{.*#+}} xmm1 = xmm0[3,3,3,3]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm1, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm1
; X64-AVX1-i64-NEXT:    vshufpd {{.*#+}} xmm3 = xmm0[1,0]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm3, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm3
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm3[0],xmm1[0]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm3
; X64-AVX1-i64-NEXT:    vmovshdup {{.*#+}} xmm0 = xmm0[1,1,3,3]
; X64-AVX1-i64-NEXT:    vcvtss2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm0
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm3[0],xmm0[0]
; X64-AVX1-i64-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm1
; X64-AVX1-i64-NEXT:    vmovaps %ymm2, %ymm0
; X64-AVX1-i64-NEXT:    retq
;
; AVX512-i64-LABEL: lrint_v8f32:
; AVX512-i64:       # %bb.0:
; AVX512-i64-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX512-i64-NEXT:    vshufps {{.*#+}} xmm2 = xmm1[3,3,3,3]
; AVX512-i64-NEXT:    vcvtss2si %xmm2, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm2
; AVX512-i64-NEXT:    vshufpd {{.*#+}} xmm3 = xmm1[1,0]
; AVX512-i64-NEXT:    vcvtss2si %xmm3, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm3
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm3[0],xmm2[0]
; AVX512-i64-NEXT:    vcvtss2si %xmm1, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm3
; AVX512-i64-NEXT:    vmovshdup {{.*#+}} xmm1 = xmm1[1,1,3,3]
; AVX512-i64-NEXT:    vcvtss2si %xmm1, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm1
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm3[0],xmm1[0]
; AVX512-i64-NEXT:    vinserti128 $1, %xmm2, %ymm1, %ymm1
; AVX512-i64-NEXT:    vshufps {{.*#+}} xmm2 = xmm0[3,3,3,3]
; AVX512-i64-NEXT:    vcvtss2si %xmm2, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm2
; AVX512-i64-NEXT:    vshufpd {{.*#+}} xmm3 = xmm0[1,0]
; AVX512-i64-NEXT:    vcvtss2si %xmm3, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm3
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm3[0],xmm2[0]
; AVX512-i64-NEXT:    vcvtss2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm3
; AVX512-i64-NEXT:    vmovshdup {{.*#+}} xmm0 = xmm0[1,1,3,3]
; AVX512-i64-NEXT:    vcvtss2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm0
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm3[0],xmm0[0]
; AVX512-i64-NEXT:    vinserti128 $1, %xmm2, %ymm0, %ymm0
; AVX512-i64-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512-i64-NEXT:    retq
;
; AVX512DQ-i64-LABEL: lrint_v8f32:
; AVX512DQ-i64:       # %bb.0:
; AVX512DQ-i64-NEXT:    vcvtps2qq %ymm0, %zmm0
; AVX512DQ-i64-NEXT:    retq
  %a = call <8 x iXLen> @llvm.lrint.v8iXLen.v8f32(<8 x float> %x)
  ret <8 x iXLen> %a
}
declare <8 x iXLen> @llvm.lrint.v8iXLen.v8f32(<8 x float>)

define <16 x iXLen> @lrint_v16iXLen_v16f32(<16 x float> %x) {
  %a = call <16 x iXLen> @llvm.lrint.v16iXLen.v16f32(<16 x float> %x)
  ret <16 x iXLen> %a
}
declare <16 x iXLen> @llvm.lrint.v16iXLen.v16f32(<16 x float>)

define <1 x iXLen> @lrint_v1f64(<1 x double> %x) {
; X86-SSE2-LABEL: lrint_v1f64:
; X86-SSE2:       # %bb.0:
; X86-SSE2-NEXT:    cvtsd2si {{[0-9]+}}(%esp), %eax
; X86-SSE2-NEXT:    retl
;
; X86-AVX-LABEL: lrint_v1f64:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    vcvtsd2si {{[0-9]+}}(%esp), %eax
; X86-AVX-NEXT:    retl
;
; X64-AVX-i32-LABEL: lrint_v1f64:
; X64-AVX-i32:       # %bb.0:
; X64-AVX-i32-NEXT:    vcvtsd2si %xmm0, %eax
; X64-AVX-i32-NEXT:    retq
;
; X64-AVX-i64-LABEL: lrint_v1f64:
; X64-AVX-i64:       # %bb.0:
; X64-AVX-i64-NEXT:    vcvtsd2si %xmm0, %rax
; X64-AVX-i64-NEXT:    retq
  %a = call <1 x iXLen> @llvm.lrint.v1iXLen.v1f64(<1 x double> %x)
  ret <1 x iXLen> %a
}
declare <1 x iXLen> @llvm.lrint.v1iXLen.v1f64(<1 x double>)

define <2 x iXLen> @lrint_v2f64(<2 x double> %x) {
; X86-SSE2-LABEL: lrint_v2f64:
; X86-SSE2:       # %bb.0:
; X86-SSE2-NEXT:    cvtsd2si %xmm0, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm1
; X86-SSE2-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1,1]
; X86-SSE2-NEXT:    cvtsd2si %xmm0, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm0
; X86-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; X86-SSE2-NEXT:    movdqa %xmm1, %xmm0
; X86-SSE2-NEXT:    retl
;
; X86-AVX-LABEL: lrint_v2f64:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    vshufpd {{.*#+}} xmm1 = xmm0[1,0]
; X86-AVX-NEXT:    vcvtsd2si %xmm1, %eax
; X86-AVX-NEXT:    vcvtsd2si %xmm0, %ecx
; X86-AVX-NEXT:    vmovd %ecx, %xmm0
; X86-AVX-NEXT:    vpinsrd $1, %eax, %xmm0, %xmm0
; X86-AVX-NEXT:    retl
;
; X64-AVX-i32-LABEL: lrint_v2f64:
; X64-AVX-i32:       # %bb.0:
; X64-AVX-i32-NEXT:    vshufpd {{.*#+}} xmm1 = xmm0[1,0]
; X64-AVX-i32-NEXT:    vcvtsd2si %xmm1, %eax
; X64-AVX-i32-NEXT:    vcvtsd2si %xmm0, %ecx
; X64-AVX-i32-NEXT:    vmovd %ecx, %xmm0
; X64-AVX-i32-NEXT:    vpinsrd $1, %eax, %xmm0, %xmm0
; X64-AVX-i32-NEXT:    retq
;
; X64-AVX1-i64-LABEL: lrint_v2f64:
; X64-AVX1-i64:       # %bb.0:
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm1
; X64-AVX1-i64-NEXT:    vshufpd {{.*#+}} xmm0 = xmm0[1,0]
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm0
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; X64-AVX1-i64-NEXT:    retq
;
; AVX512-i64-LABEL: lrint_v2f64:
; AVX512-i64:       # %bb.0:
; AVX512-i64-NEXT:    vcvtsd2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm1
; AVX512-i64-NEXT:    vshufpd {{.*#+}} xmm0 = xmm0[1,0]
; AVX512-i64-NEXT:    vcvtsd2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm0
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX512-i64-NEXT:    retq
;
; AVX512DQ-i64-LABEL: lrint_v2f64:
; AVX512DQ-i64:       # %bb.0:
; AVX512DQ-i64-NEXT:    vcvtpd2qq %xmm0, %xmm0
; AVX512DQ-i64-NEXT:    retq
  %a = call <2 x iXLen> @llvm.lrint.v2iXLen.v2f64(<2 x double> %x)
  ret <2 x iXLen> %a
}
declare <2 x iXLen> @llvm.lrint.v2iXLen.v2f64(<2 x double>)

define <4 x iXLen> @lrint_v4f64(<4 x double> %x) {
; X86-SSE2-LABEL: lrint_v4f64:
; X86-SSE2:       # %bb.0:
; X86-SSE2-NEXT:    cvtsd2si %xmm1, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm2
; X86-SSE2-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1,1]
; X86-SSE2-NEXT:    cvtsd2si %xmm1, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm1
; X86-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
; X86-SSE2-NEXT:    cvtsd2si %xmm0, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm1
; X86-SSE2-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1,1]
; X86-SSE2-NEXT:    cvtsd2si %xmm0, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm0
; X86-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; X86-SSE2-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
; X86-SSE2-NEXT:    movdqa %xmm1, %xmm0
; X86-SSE2-NEXT:    retl
;
; X86-AVX-LABEL: lrint_v4f64:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    vcvtpd2dq %ymm0, %xmm0
; X86-AVX-NEXT:    vzeroupper
; X86-AVX-NEXT:    retl
;
; X64-AVX-i32-LABEL: lrint_v4f64:
; X64-AVX-i32:       # %bb.0:
; X64-AVX-i32-NEXT:    vcvtpd2dq %ymm0, %xmm0
; X64-AVX-i32-NEXT:    vzeroupper
; X64-AVX-i32-NEXT:    retq
;
; X64-AVX1-i64-LABEL: lrint_v4f64:
; X64-AVX1-i64:       # %bb.0:
; X64-AVX1-i64-NEXT:    vextractf128 $1, %ymm0, %xmm1
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm1, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm2
; X64-AVX1-i64-NEXT:    vshufpd {{.*#+}} xmm1 = xmm1[1,0]
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm1, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm1
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm2
; X64-AVX1-i64-NEXT:    vshufpd {{.*#+}} xmm0 = xmm0[1,0]
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm0
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm2[0],xmm0[0]
; X64-AVX1-i64-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; X64-AVX1-i64-NEXT:    retq
;
; AVX512-i64-LABEL: lrint_v4f64:
; AVX512-i64:       # %bb.0:
; AVX512-i64-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX512-i64-NEXT:    vcvtsd2si %xmm1, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm2
; AVX512-i64-NEXT:    vshufpd {{.*#+}} xmm1 = xmm1[1,0]
; AVX512-i64-NEXT:    vcvtsd2si %xmm1, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm1
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
; AVX512-i64-NEXT:    vcvtsd2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm2
; AVX512-i64-NEXT:    vshufpd {{.*#+}} xmm0 = xmm0[1,0]
; AVX512-i64-NEXT:    vcvtsd2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm0
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm2[0],xmm0[0]
; AVX512-i64-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX512-i64-NEXT:    retq
;
; AVX512DQ-i64-LABEL: lrint_v4f64:
; AVX512DQ-i64:       # %bb.0:
; AVX512DQ-i64-NEXT:    vcvtpd2qq %ymm0, %ymm0
; AVX512DQ-i64-NEXT:    retq
  %a = call <4 x iXLen> @llvm.lrint.v4iXLen.v4f64(<4 x double> %x)
  ret <4 x iXLen> %a
}
declare <4 x iXLen> @llvm.lrint.v4iXLen.v4f64(<4 x double>)

define <8 x iXLen> @lrint_v8f64(<8 x double> %x) {
; X86-SSE2-LABEL: lrint_v8f64:
; X86-SSE2:       # %bb.0:
; X86-SSE2-NEXT:    pushl %ebp
; X86-SSE2-NEXT:    .cfi_def_cfa_offset 8
; X86-SSE2-NEXT:    .cfi_offset %ebp, -8
; X86-SSE2-NEXT:    movl %esp, %ebp
; X86-SSE2-NEXT:    .cfi_def_cfa_register %ebp
; X86-SSE2-NEXT:    andl $-16, %esp
; X86-SSE2-NEXT:    subl $16, %esp
; X86-SSE2-NEXT:    movapd %xmm0, %xmm3
; X86-SSE2-NEXT:    movapd 8(%ebp), %xmm4
; X86-SSE2-NEXT:    cvtsd2si %xmm1, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm5
; X86-SSE2-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1,1]
; X86-SSE2-NEXT:    cvtsd2si %xmm1, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm0
; X86-SSE2-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1]
; X86-SSE2-NEXT:    cvtsd2si %xmm3, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm0
; X86-SSE2-NEXT:    unpckhpd {{.*#+}} xmm3 = xmm3[1,1]
; X86-SSE2-NEXT:    cvtsd2si %xmm3, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm1
; X86-SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X86-SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm5[0]
; X86-SSE2-NEXT:    cvtsd2si %xmm4, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm3
; X86-SSE2-NEXT:    unpckhpd {{.*#+}} xmm4 = xmm4[1,1]
; X86-SSE2-NEXT:    cvtsd2si %xmm4, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm1
; X86-SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
; X86-SSE2-NEXT:    cvtsd2si %xmm2, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm1
; X86-SSE2-NEXT:    unpckhpd {{.*#+}} xmm2 = xmm2[1,1]
; X86-SSE2-NEXT:    cvtsd2si %xmm2, %eax
; X86-SSE2-NEXT:    movd %eax, %xmm2
; X86-SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
; X86-SSE2-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm3[0]
; X86-SSE2-NEXT:    movl %ebp, %esp
; X86-SSE2-NEXT:    popl %ebp
; X86-SSE2-NEXT:    .cfi_def_cfa %esp, 4
; X86-SSE2-NEXT:    retl
;
; X86-AVX1-LABEL: lrint_v8f64:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    vcvtpd2dq %ymm0, %xmm0
; X86-AVX1-NEXT:    vcvtpd2dq %ymm1, %xmm1
; X86-AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; X86-AVX1-NEXT:    retl
;
; AVX512-i32-LABEL: lrint_v8f64:
; AVX512-i32:       # %bb.0:
; AVX512-i32-NEXT:    vcvtpd2dq %zmm0, %ymm0
; AVX512-i32-NEXT:    ret{{[l|q]}}
;
; X64-AVX1-i32-LABEL: lrint_v8f64:
; X64-AVX1-i32:       # %bb.0:
; X64-AVX1-i32-NEXT:    vcvtpd2dq %ymm0, %xmm0
; X64-AVX1-i32-NEXT:    vcvtpd2dq %ymm1, %xmm1
; X64-AVX1-i32-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; X64-AVX1-i32-NEXT:    retq
;
; X64-AVX1-i64-LABEL: lrint_v8f64:
; X64-AVX1-i64:       # %bb.0:
; X64-AVX1-i64-NEXT:    vextractf128 $1, %ymm0, %xmm2
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm2, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm3
; X64-AVX1-i64-NEXT:    vshufpd {{.*#+}} xmm2 = xmm2[1,0]
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm2, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm2
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm3[0],xmm2[0]
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm3
; X64-AVX1-i64-NEXT:    vshufpd {{.*#+}} xmm0 = xmm0[1,0]
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm0, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm0
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm3[0],xmm0[0]
; X64-AVX1-i64-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
; X64-AVX1-i64-NEXT:    vextractf128 $1, %ymm1, %xmm2
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm2, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm3
; X64-AVX1-i64-NEXT:    vshufpd {{.*#+}} xmm2 = xmm2[1,0]
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm2, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm2
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm3[0],xmm2[0]
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm1, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm3
; X64-AVX1-i64-NEXT:    vshufpd {{.*#+}} xmm1 = xmm1[1,0]
; X64-AVX1-i64-NEXT:    vcvtsd2si %xmm1, %rax
; X64-AVX1-i64-NEXT:    vmovq %rax, %xmm1
; X64-AVX1-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm3[0],xmm1[0]
; X64-AVX1-i64-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm1
; X64-AVX1-i64-NEXT:    retq
;
; AVX512-i64-LABEL: lrint_v8f64:
; AVX512-i64:       # %bb.0:
; AVX512-i64-NEXT:    vextractf32x4 $3, %zmm0, %xmm1
; AVX512-i64-NEXT:    vcvtsd2si %xmm1, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm2
; AVX512-i64-NEXT:    vshufpd {{.*#+}} xmm1 = xmm1[1,0]
; AVX512-i64-NEXT:    vcvtsd2si %xmm1, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm1
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
; AVX512-i64-NEXT:    vextractf32x4 $2, %zmm0, %xmm2
; AVX512-i64-NEXT:    vcvtsd2si %xmm2, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm3
; AVX512-i64-NEXT:    vshufpd {{.*#+}} xmm2 = xmm2[1,0]
; AVX512-i64-NEXT:    vcvtsd2si %xmm2, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm2
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm3[0],xmm2[0]
; AVX512-i64-NEXT:    vinserti128 $1, %xmm1, %ymm2, %ymm1
; AVX512-i64-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX512-i64-NEXT:    vcvtsd2si %xmm2, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm3
; AVX512-i64-NEXT:    vshufpd {{.*#+}} xmm2 = xmm2[1,0]
; AVX512-i64-NEXT:    vcvtsd2si %xmm2, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm2
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm3[0],xmm2[0]
; AVX512-i64-NEXT:    vcvtsd2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm3
; AVX512-i64-NEXT:    vshufpd {{.*#+}} xmm0 = xmm0[1,0]
; AVX512-i64-NEXT:    vcvtsd2si %xmm0, %rax
; AVX512-i64-NEXT:    vmovq %rax, %xmm0
; AVX512-i64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm3[0],xmm0[0]
; AVX512-i64-NEXT:    vinserti128 $1, %xmm2, %ymm0, %ymm0
; AVX512-i64-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512-i64-NEXT:    retq
;
; AVX512DQ-i64-LABEL: lrint_v8f64:
; AVX512DQ-i64:       # %bb.0:
; AVX512DQ-i64-NEXT:    vcvtpd2qq %zmm0, %zmm0
; AVX512DQ-i64-NEXT:    retq
  %a = call <8 x iXLen> @llvm.lrint.v8iXLen.v8f64(<8 x double> %x)
  ret <8 x iXLen> %a
}
declare <8 x iXLen> @llvm.lrint.v8iXLen.v8f64(<8 x double>)