; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
; RUN: -mcpu=pwr8 -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr < %s | \
; RUN: FileCheck %s --check-prefix=CHECK-P8
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
; RUN: -mcpu=pwr9 -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr < %s | \
; RUN: FileCheck %s --check-prefix=CHECK-P9
; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu \
; RUN: -mcpu=pwr9 -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr < %s | \
; RUN: FileCheck %s --check-prefix=CHECK-BE
define <2 x double> @test2elt(i16 %a.coerce) local_unnamed_addr #0 {
; CHECK-P8-LABEL: test2elt:
; CHECK-P8: # %bb.0: # %entry
; CHECK-P8-NEXT: addis r4, r2, .LCPI0_0@toc@ha
; CHECK-P8-NEXT: mtvsrd v3, r3
; CHECK-P8-NEXT: xxlxor v4, v4, v4
; CHECK-P8-NEXT: addi r4, r4, .LCPI0_0@toc@l
; CHECK-P8-NEXT: lxvd2x vs0, 0, r4
; CHECK-P8-NEXT: xxswapd v2, vs0
; CHECK-P8-NEXT: vperm v2, v4, v3, v2
; CHECK-P8-NEXT: xvcvuxddp v2, v2
; CHECK-P8-NEXT: blr
;
; CHECK-P9-LABEL: test2elt:
; CHECK-P9: # %bb.0: # %entry
; CHECK-P9-NEXT: mtfprd f0, r3
; CHECK-P9-NEXT: addis r3, r2, .LCPI0_0@toc@ha
; CHECK-P9-NEXT: xxlxor vs2, vs2, vs2
; CHECK-P9-NEXT: addi r3, r3, .LCPI0_0@toc@l
; CHECK-P9-NEXT: lxv vs1, 0(r3)
; CHECK-P9-NEXT: xxperm vs0, vs2, vs1
; CHECK-P9-NEXT: xvcvuxddp v2, vs0
; CHECK-P9-NEXT: blr
;
; CHECK-BE-LABEL: test2elt:
; CHECK-BE: # %bb.0: # %entry
; CHECK-BE-NEXT: mtfprwz f0, r3
; CHECK-BE-NEXT: addis r3, r2, .LCPI0_0@toc@ha
; CHECK-BE-NEXT: xxlxor vs2, vs2, vs2
; CHECK-BE-NEXT: addi r3, r3, .LCPI0_0@toc@l
; CHECK-BE-NEXT: lxv vs1, 0(r3)
; CHECK-BE-NEXT: xxperm vs0, vs2, vs1
; CHECK-BE-NEXT: xvcvuxddp v2, vs0
; CHECK-BE-NEXT: blr
entry:
%0 = bitcast i16 %a.coerce to <2 x i8>
%1 = uitofp <2 x i8> %0 to <2 x double>
ret <2 x double> %1
}
define void @test4elt(ptr noalias nocapture sret(<4 x double>) %agg.result, i32 %a.coerce) local_unnamed_addr #1 {
; CHECK-P8-LABEL: test4elt:
; CHECK-P8: # %bb.0: # %entry
; CHECK-P8-NEXT: addis r5, r2, .LCPI1_0@toc@ha
; CHECK-P8-NEXT: mtvsrwz v4, r4
; CHECK-P8-NEXT: xxlxor v5, v5, v5
; CHECK-P8-NEXT: li r4, 16
; CHECK-P8-NEXT: addi r5, r5, .LCPI1_0@toc@l
; CHECK-P8-NEXT: lxvd2x vs0, 0, r5
; CHECK-P8-NEXT: addis r5, r2, .LCPI1_1@toc@ha
; CHECK-P8-NEXT: addi r5, r5, .LCPI1_1@toc@l
; CHECK-P8-NEXT: lxvd2x vs1, 0, r5
; CHECK-P8-NEXT: xxswapd v2, vs0
; CHECK-P8-NEXT: vperm v2, v5, v4, v2
; CHECK-P8-NEXT: xvcvuxddp vs0, v2
; CHECK-P8-NEXT: xxswapd v3, vs1
; CHECK-P8-NEXT: vperm v3, v5, v4, v3
; CHECK-P8-NEXT: xvcvuxddp vs1, v3
; CHECK-P8-NEXT: xxswapd vs0, vs0
; CHECK-P8-NEXT: xxswapd vs1, vs1
; CHECK-P8-NEXT: stxvd2x vs1, r3, r4
; CHECK-P8-NEXT: stxvd2x vs0, 0, r3
; CHECK-P8-NEXT: blr
;
; CHECK-P9-LABEL: test4elt:
; CHECK-P9: # %bb.0: # %entry
; CHECK-P9-NEXT: mtvsrwz v2, r4
; CHECK-P9-NEXT: addis r4, r2, .LCPI1_0@toc@ha
; CHECK-P9-NEXT: xxlxor v4, v4, v4
; CHECK-P9-NEXT: addi r4, r4, .LCPI1_0@toc@l
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI1_1@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI1_1@toc@l
; CHECK-P9-NEXT: vperm v3, v4, v2, v3
; CHECK-P9-NEXT: xvcvuxddp vs0, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: vperm v2, v4, v2, v3
; CHECK-P9-NEXT: stxv vs0, 0(r3)
; CHECK-P9-NEXT: xvcvuxddp vs1, v2
; CHECK-P9-NEXT: stxv vs1, 16(r3)
; CHECK-P9-NEXT: blr
;
; CHECK-BE-LABEL: test4elt:
; CHECK-BE: # %bb.0: # %entry
; CHECK-BE-NEXT: mtvsrwz v2, r4
; CHECK-BE-NEXT: addis r4, r2, .LCPI1_0@toc@ha
; CHECK-BE-NEXT: xxlxor v4, v4, v4
; CHECK-BE-NEXT: addi r4, r4, .LCPI1_0@toc@l
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI1_1@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI1_1@toc@l
; CHECK-BE-NEXT: vperm v3, v4, v2, v3
; CHECK-BE-NEXT: xvcvuxddp vs0, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: vperm v2, v4, v2, v3
; CHECK-BE-NEXT: stxv vs0, 0(r3)
; CHECK-BE-NEXT: xvcvuxddp vs1, v2
; CHECK-BE-NEXT: stxv vs1, 16(r3)
; CHECK-BE-NEXT: blr
entry:
%0 = bitcast i32 %a.coerce to <4 x i8>
%1 = uitofp <4 x i8> %0 to <4 x double>
store <4 x double> %1, ptr %agg.result, align 32
ret void
}
define void @test8elt(ptr noalias nocapture sret(<8 x double>) %agg.result, i64 %a.coerce) local_unnamed_addr #1 {
; CHECK-P8-LABEL: test8elt:
; CHECK-P8: # %bb.0: # %entry
; CHECK-P8-NEXT: addis r5, r2, .LCPI2_0@toc@ha
; CHECK-P8-NEXT: mtvsrd v0, r4
; CHECK-P8-NEXT: xxlxor v1, v1, v1
; CHECK-P8-NEXT: li r4, 48
; CHECK-P8-NEXT: addi r5, r5, .LCPI2_0@toc@l
; CHECK-P8-NEXT: lxvd2x vs0, 0, r5
; CHECK-P8-NEXT: addis r5, r2, .LCPI2_1@toc@ha
; CHECK-P8-NEXT: addi r5, r5, .LCPI2_1@toc@l
; CHECK-P8-NEXT: lxvd2x vs1, 0, r5
; CHECK-P8-NEXT: addis r5, r2, .LCPI2_3@toc@ha
; CHECK-P8-NEXT: addi r5, r5, .LCPI2_3@toc@l
; CHECK-P8-NEXT: lxvd2x vs2, 0, r5
; CHECK-P8-NEXT: addis r5, r2, .LCPI2_2@toc@ha
; CHECK-P8-NEXT: addi r5, r5, .LCPI2_2@toc@l
; CHECK-P8-NEXT: xxswapd v3, vs0
; CHECK-P8-NEXT: vperm v3, v1, v0, v3
; CHECK-P8-NEXT: xvcvuxddp vs0, v3
; CHECK-P8-NEXT: xxswapd v4, vs1
; CHECK-P8-NEXT: vperm v4, v1, v0, v4
; CHECK-P8-NEXT: xvcvuxddp vs1, v4
; CHECK-P8-NEXT: xxswapd v2, vs2
; CHECK-P8-NEXT: lxvd2x vs2, 0, r5
; CHECK-P8-NEXT: vperm v2, v1, v0, v2
; CHECK-P8-NEXT: xxswapd vs0, vs0
; CHECK-P8-NEXT: xxswapd v5, vs2
; CHECK-P8-NEXT: xvcvuxddp vs2, v2
; CHECK-P8-NEXT: xxswapd vs1, vs1
; CHECK-P8-NEXT: vperm v5, v1, v0, v5
; CHECK-P8-NEXT: xvcvuxddp vs3, v5
; CHECK-P8-NEXT: xxswapd vs2, vs2
; CHECK-P8-NEXT: stxvd2x vs2, r3, r4
; CHECK-P8-NEXT: li r4, 32
; CHECK-P8-NEXT: xxswapd vs3, vs3
; CHECK-P8-NEXT: stxvd2x vs3, r3, r4
; CHECK-P8-NEXT: li r4, 16
; CHECK-P8-NEXT: stxvd2x vs1, r3, r4
; CHECK-P8-NEXT: stxvd2x vs0, 0, r3
; CHECK-P8-NEXT: blr
;
; CHECK-P9-LABEL: test8elt:
; CHECK-P9: # %bb.0: # %entry
; CHECK-P9-NEXT: mtvsrd v2, r4
; CHECK-P9-NEXT: addis r4, r2, .LCPI2_0@toc@ha
; CHECK-P9-NEXT: xxlxor v4, v4, v4
; CHECK-P9-NEXT: addi r4, r4, .LCPI2_0@toc@l
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI2_1@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI2_1@toc@l
; CHECK-P9-NEXT: vperm v3, v4, v2, v3
; CHECK-P9-NEXT: xvcvuxddp vs0, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI2_2@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI2_2@toc@l
; CHECK-P9-NEXT: vperm v3, v4, v2, v3
; CHECK-P9-NEXT: stxv vs0, 0(r3)
; CHECK-P9-NEXT: xvcvuxddp vs1, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI2_3@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI2_3@toc@l
; CHECK-P9-NEXT: vperm v3, v4, v2, v3
; CHECK-P9-NEXT: stxv vs1, 16(r3)
; CHECK-P9-NEXT: xvcvuxddp vs2, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: vperm v2, v4, v2, v3
; CHECK-P9-NEXT: stxv vs2, 32(r3)
; CHECK-P9-NEXT: xvcvuxddp vs3, v2
; CHECK-P9-NEXT: stxv vs3, 48(r3)
; CHECK-P9-NEXT: blr
;
; CHECK-BE-LABEL: test8elt:
; CHECK-BE: # %bb.0: # %entry
; CHECK-BE-NEXT: mtvsrd v2, r4
; CHECK-BE-NEXT: addis r4, r2, .LCPI2_0@toc@ha
; CHECK-BE-NEXT: xxlxor v4, v4, v4
; CHECK-BE-NEXT: addi r4, r4, .LCPI2_0@toc@l
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI2_1@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI2_1@toc@l
; CHECK-BE-NEXT: vperm v3, v4, v2, v3
; CHECK-BE-NEXT: xvcvuxddp vs0, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI2_2@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI2_2@toc@l
; CHECK-BE-NEXT: vperm v3, v4, v2, v3
; CHECK-BE-NEXT: stxv vs0, 0(r3)
; CHECK-BE-NEXT: xvcvuxddp vs1, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI2_3@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI2_3@toc@l
; CHECK-BE-NEXT: vperm v3, v4, v2, v3
; CHECK-BE-NEXT: stxv vs1, 16(r3)
; CHECK-BE-NEXT: xvcvuxddp vs2, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: vperm v2, v4, v2, v3
; CHECK-BE-NEXT: stxv vs2, 32(r3)
; CHECK-BE-NEXT: xvcvuxddp vs3, v2
; CHECK-BE-NEXT: stxv vs3, 48(r3)
; CHECK-BE-NEXT: blr
entry:
%0 = bitcast i64 %a.coerce to <8 x i8>
%1 = uitofp <8 x i8> %0 to <8 x double>
store <8 x double> %1, ptr %agg.result, align 64
ret void
}
define void @test16elt(ptr noalias nocapture sret(<16 x double>) %agg.result, <16 x i8> %a) local_unnamed_addr #2 {
; CHECK-P8-LABEL: test16elt:
; CHECK-P8: # %bb.0: # %entry
; CHECK-P8-NEXT: addis r4, r2, .LCPI3_0@toc@ha
; CHECK-P8-NEXT: xxlxor v0, v0, v0
; CHECK-P8-NEXT: addi r4, r4, .LCPI3_0@toc@l
; CHECK-P8-NEXT: lxvd2x vs0, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI3_1@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI3_1@toc@l
; CHECK-P8-NEXT: lxvd2x vs1, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI3_3@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI3_3@toc@l
; CHECK-P8-NEXT: lxvd2x vs2, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI3_4@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI3_4@toc@l
; CHECK-P8-NEXT: lxvd2x vs3, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI3_6@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI3_6@toc@l
; CHECK-P8-NEXT: lxvd2x vs4, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI3_7@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI3_7@toc@l
; CHECK-P8-NEXT: lxvd2x vs5, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI3_5@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI3_5@toc@l
; CHECK-P8-NEXT: xxswapd v7, vs0
; CHECK-P8-NEXT: vperm v7, v0, v2, v7
; CHECK-P8-NEXT: xvcvuxddp vs0, v7
; CHECK-P8-NEXT: xxswapd v8, vs1
; CHECK-P8-NEXT: vperm v8, v0, v2, v8
; CHECK-P8-NEXT: xvcvuxddp vs1, v8
; CHECK-P8-NEXT: xxswapd v1, vs2
; CHECK-P8-NEXT: vperm v1, v0, v2, v1
; CHECK-P8-NEXT: xxswapd vs0, vs0
; CHECK-P8-NEXT: xxswapd v6, vs3
; CHECK-P8-NEXT: vperm v6, v0, v2, v6
; CHECK-P8-NEXT: xvcvuxddp vs6, v6
; CHECK-P8-NEXT: xxswapd vs1, vs1
; CHECK-P8-NEXT: xxswapd v3, vs4
; CHECK-P8-NEXT: lxvd2x vs4, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI3_2@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI3_2@toc@l
; CHECK-P8-NEXT: lxvd2x vs2, 0, r4
; CHECK-P8-NEXT: li r4, 112
; CHECK-P8-NEXT: vperm v3, v0, v2, v3
; CHECK-P8-NEXT: xvcvuxddp vs3, v3
; CHECK-P8-NEXT: xxswapd v4, vs5
; CHECK-P8-NEXT: xvcvuxddp vs5, v1
; CHECK-P8-NEXT: vperm v4, v0, v2, v4
; CHECK-P8-NEXT: xxswapd v5, vs4
; CHECK-P8-NEXT: xvcvuxddp vs4, v4
; CHECK-P8-NEXT: vperm v5, v0, v2, v5
; CHECK-P8-NEXT: xxswapd vs3, vs3
; CHECK-P8-NEXT: xvcvuxddp vs7, v5
; CHECK-P8-NEXT: xxswapd v9, vs2
; CHECK-P8-NEXT: xxswapd vs4, vs4
; CHECK-P8-NEXT: vperm v2, v0, v2, v9
; CHECK-P8-NEXT: stxvd2x vs4, r3, r4
; CHECK-P8-NEXT: li r4, 96
; CHECK-P8-NEXT: xvcvuxddp vs2, v2
; CHECK-P8-NEXT: stxvd2x vs3, r3, r4
; CHECK-P8-NEXT: li r4, 80
; CHECK-P8-NEXT: xxswapd vs4, vs5
; CHECK-P8-NEXT: xxswapd vs7, vs7
; CHECK-P8-NEXT: xxswapd vs3, vs6
; CHECK-P8-NEXT: stxvd2x vs7, r3, r4
; CHECK-P8-NEXT: li r4, 64
; CHECK-P8-NEXT: stxvd2x vs3, r3, r4
; CHECK-P8-NEXT: li r4, 48
; CHECK-P8-NEXT: stxvd2x vs4, r3, r4
; CHECK-P8-NEXT: li r4, 32
; CHECK-P8-NEXT: xxswapd vs2, vs2
; CHECK-P8-NEXT: stxvd2x vs2, r3, r4
; CHECK-P8-NEXT: li r4, 16
; CHECK-P8-NEXT: stxvd2x vs1, r3, r4
; CHECK-P8-NEXT: stxvd2x vs0, 0, r3
; CHECK-P8-NEXT: blr
;
; CHECK-P9-LABEL: test16elt:
; CHECK-P9: # %bb.0: # %entry
; CHECK-P9-NEXT: addis r4, r2, .LCPI3_0@toc@ha
; CHECK-P9-NEXT: xxlxor v4, v4, v4
; CHECK-P9-NEXT: addi r4, r4, .LCPI3_0@toc@l
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI3_1@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI3_1@toc@l
; CHECK-P9-NEXT: vperm v3, v4, v2, v3
; CHECK-P9-NEXT: xvcvuxddp vs0, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI3_2@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI3_2@toc@l
; CHECK-P9-NEXT: vperm v3, v4, v2, v3
; CHECK-P9-NEXT: stxv vs0, 0(r3)
; CHECK-P9-NEXT: xvcvuxddp vs1, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI3_3@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI3_3@toc@l
; CHECK-P9-NEXT: vperm v3, v4, v2, v3
; CHECK-P9-NEXT: stxv vs1, 16(r3)
; CHECK-P9-NEXT: xvcvuxddp vs2, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI3_4@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI3_4@toc@l
; CHECK-P9-NEXT: vperm v3, v4, v2, v3
; CHECK-P9-NEXT: stxv vs2, 32(r3)
; CHECK-P9-NEXT: xvcvuxddp vs3, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI3_5@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI3_5@toc@l
; CHECK-P9-NEXT: vperm v3, v4, v2, v3
; CHECK-P9-NEXT: stxv vs3, 48(r3)
; CHECK-P9-NEXT: xvcvuxddp vs4, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI3_6@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI3_6@toc@l
; CHECK-P9-NEXT: vperm v3, v4, v2, v3
; CHECK-P9-NEXT: stxv vs4, 64(r3)
; CHECK-P9-NEXT: xvcvuxddp vs5, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI3_7@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI3_7@toc@l
; CHECK-P9-NEXT: vperm v3, v4, v2, v3
; CHECK-P9-NEXT: stxv vs5, 80(r3)
; CHECK-P9-NEXT: xvcvuxddp vs6, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: vperm v2, v4, v2, v3
; CHECK-P9-NEXT: stxv vs6, 96(r3)
; CHECK-P9-NEXT: xvcvuxddp vs7, v2
; CHECK-P9-NEXT: stxv vs7, 112(r3)
; CHECK-P9-NEXT: blr
;
; CHECK-BE-LABEL: test16elt:
; CHECK-BE: # %bb.0: # %entry
; CHECK-BE-NEXT: addis r4, r2, .LCPI3_0@toc@ha
; CHECK-BE-NEXT: xxlxor v4, v4, v4
; CHECK-BE-NEXT: addi r4, r4, .LCPI3_0@toc@l
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI3_1@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI3_1@toc@l
; CHECK-BE-NEXT: vperm v3, v4, v2, v3
; CHECK-BE-NEXT: xvcvuxddp vs0, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI3_2@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI3_2@toc@l
; CHECK-BE-NEXT: vperm v3, v4, v2, v3
; CHECK-BE-NEXT: stxv vs0, 0(r3)
; CHECK-BE-NEXT: xvcvuxddp vs1, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI3_3@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI3_3@toc@l
; CHECK-BE-NEXT: vperm v3, v4, v2, v3
; CHECK-BE-NEXT: stxv vs1, 16(r3)
; CHECK-BE-NEXT: xvcvuxddp vs2, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI3_4@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI3_4@toc@l
; CHECK-BE-NEXT: vperm v3, v4, v2, v3
; CHECK-BE-NEXT: stxv vs2, 32(r3)
; CHECK-BE-NEXT: xvcvuxddp vs3, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI3_5@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI3_5@toc@l
; CHECK-BE-NEXT: vperm v3, v4, v2, v3
; CHECK-BE-NEXT: stxv vs3, 48(r3)
; CHECK-BE-NEXT: xvcvuxddp vs4, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI3_6@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI3_6@toc@l
; CHECK-BE-NEXT: vperm v3, v4, v2, v3
; CHECK-BE-NEXT: stxv vs4, 64(r3)
; CHECK-BE-NEXT: xvcvuxddp vs5, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI3_7@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI3_7@toc@l
; CHECK-BE-NEXT: vperm v3, v4, v2, v3
; CHECK-BE-NEXT: stxv vs5, 80(r3)
; CHECK-BE-NEXT: xvcvuxddp vs6, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: vperm v2, v4, v2, v3
; CHECK-BE-NEXT: stxv vs6, 96(r3)
; CHECK-BE-NEXT: xvcvuxddp vs7, v2
; CHECK-BE-NEXT: stxv vs7, 112(r3)
; CHECK-BE-NEXT: blr
entry:
%0 = uitofp <16 x i8> %a to <16 x double>
store <16 x double> %0, ptr %agg.result, align 128
ret void
}
define <2 x double> @test2elt_signed(i16 %a.coerce) local_unnamed_addr #0 {
; CHECK-P8-LABEL: test2elt_signed:
; CHECK-P8: # %bb.0: # %entry
; CHECK-P8-NEXT: addis r4, r2, .LCPI4_0@toc@ha
; CHECK-P8-NEXT: mtvsrd v3, r3
; CHECK-P8-NEXT: addis r3, r2, .LCPI4_1@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI4_0@toc@l
; CHECK-P8-NEXT: addi r3, r3, .LCPI4_1@toc@l
; CHECK-P8-NEXT: lxvd2x vs0, 0, r4
; CHECK-P8-NEXT: xxswapd v2, vs0
; CHECK-P8-NEXT: vperm v2, v3, v3, v2
; CHECK-P8-NEXT: lxvd2x v3, 0, r3
; CHECK-P8-NEXT: vsld v2, v2, v3
; CHECK-P8-NEXT: vsrad v2, v2, v3
; CHECK-P8-NEXT: xvcvsxddp v2, v2
; CHECK-P8-NEXT: blr
;
; CHECK-P9-LABEL: test2elt_signed:
; CHECK-P9: # %bb.0: # %entry
; CHECK-P9-NEXT: mtvsrd v2, r3
; CHECK-P9-NEXT: addis r3, r2, .LCPI4_0@toc@ha
; CHECK-P9-NEXT: addi r3, r3, .LCPI4_0@toc@l
; CHECK-P9-NEXT: lxv vs0, 0(r3)
; CHECK-P9-NEXT: xxperm v2, v2, vs0
; CHECK-P9-NEXT: vextsb2d v2, v2
; CHECK-P9-NEXT: xvcvsxddp v2, v2
; CHECK-P9-NEXT: blr
;
; CHECK-BE-LABEL: test2elt_signed:
; CHECK-BE: # %bb.0: # %entry
; CHECK-BE-NEXT: mtvsrwz v2, r3
; CHECK-BE-NEXT: addis r3, r2, .LCPI4_0@toc@ha
; CHECK-BE-NEXT: addi r3, r3, .LCPI4_0@toc@l
; CHECK-BE-NEXT: lxv vs0, 0(r3)
; CHECK-BE-NEXT: xxperm v2, v2, vs0
; CHECK-BE-NEXT: vextsb2d v2, v2
; CHECK-BE-NEXT: xvcvsxddp v2, v2
; CHECK-BE-NEXT: blr
entry:
%0 = bitcast i16 %a.coerce to <2 x i8>
%1 = sitofp <2 x i8> %0 to <2 x double>
ret <2 x double> %1
}
define void @test4elt_signed(ptr noalias nocapture sret(<4 x double>) %agg.result, i32 %a.coerce) local_unnamed_addr #1 {
; CHECK-P8-LABEL: test4elt_signed:
; CHECK-P8: # %bb.0: # %entry
; CHECK-P8-NEXT: addis r5, r2, .LCPI5_0@toc@ha
; CHECK-P8-NEXT: mtvsrwz v4, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI5_1@toc@ha
; CHECK-P8-NEXT: addi r5, r5, .LCPI5_0@toc@l
; CHECK-P8-NEXT: addi r4, r4, .LCPI5_1@toc@l
; CHECK-P8-NEXT: lxvd2x vs0, 0, r5
; CHECK-P8-NEXT: addis r5, r2, .LCPI5_2@toc@ha
; CHECK-P8-NEXT: addi r5, r5, .LCPI5_2@toc@l
; CHECK-P8-NEXT: lxvd2x vs1, 0, r5
; CHECK-P8-NEXT: xxswapd v2, vs0
; CHECK-P8-NEXT: vperm v2, v4, v4, v2
; CHECK-P8-NEXT: xxswapd v3, vs1
; CHECK-P8-NEXT: vperm v3, v4, v4, v3
; CHECK-P8-NEXT: lxvd2x v4, 0, r4
; CHECK-P8-NEXT: li r4, 16
; CHECK-P8-NEXT: vsld v2, v2, v4
; CHECK-P8-NEXT: vsrad v2, v2, v4
; CHECK-P8-NEXT: xvcvsxddp vs0, v2
; CHECK-P8-NEXT: vsld v2, v3, v4
; CHECK-P8-NEXT: vsrad v2, v2, v4
; CHECK-P8-NEXT: xvcvsxddp vs1, v2
; CHECK-P8-NEXT: xxswapd vs0, vs0
; CHECK-P8-NEXT: xxswapd vs1, vs1
; CHECK-P8-NEXT: stxvd2x vs1, r3, r4
; CHECK-P8-NEXT: stxvd2x vs0, 0, r3
; CHECK-P8-NEXT: blr
;
; CHECK-P9-LABEL: test4elt_signed:
; CHECK-P9: # %bb.0: # %entry
; CHECK-P9-NEXT: mtvsrwz v2, r4
; CHECK-P9-NEXT: addis r4, r2, .LCPI5_0@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI5_0@toc@l
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI5_1@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI5_1@toc@l
; CHECK-P9-NEXT: vperm v3, v2, v2, v3
; CHECK-P9-NEXT: vextsb2d v3, v3
; CHECK-P9-NEXT: xvcvsxddp vs0, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: vperm v2, v2, v2, v3
; CHECK-P9-NEXT: stxv vs0, 0(r3)
; CHECK-P9-NEXT: vextsb2d v2, v2
; CHECK-P9-NEXT: xvcvsxddp vs1, v2
; CHECK-P9-NEXT: stxv vs1, 16(r3)
; CHECK-P9-NEXT: blr
;
; CHECK-BE-LABEL: test4elt_signed:
; CHECK-BE: # %bb.0: # %entry
; CHECK-BE-NEXT: mtvsrwz v2, r4
; CHECK-BE-NEXT: addis r4, r2, .LCPI5_0@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI5_0@toc@l
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI5_1@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI5_1@toc@l
; CHECK-BE-NEXT: vperm v3, v2, v2, v3
; CHECK-BE-NEXT: vextsb2d v3, v3
; CHECK-BE-NEXT: xvcvsxddp vs0, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: vperm v2, v2, v2, v3
; CHECK-BE-NEXT: stxv vs0, 0(r3)
; CHECK-BE-NEXT: vextsb2d v2, v2
; CHECK-BE-NEXT: xvcvsxddp vs1, v2
; CHECK-BE-NEXT: stxv vs1, 16(r3)
; CHECK-BE-NEXT: blr
entry:
%0 = bitcast i32 %a.coerce to <4 x i8>
%1 = sitofp <4 x i8> %0 to <4 x double>
store <4 x double> %1, ptr %agg.result, align 32
ret void
}
define void @test8elt_signed(ptr noalias nocapture sret(<8 x double>) %agg.result, i64 %a.coerce) local_unnamed_addr #1 {
; CHECK-P8-LABEL: test8elt_signed:
; CHECK-P8: # %bb.0: # %entry
; CHECK-P8-NEXT: addis r5, r2, .LCPI6_0@toc@ha
; CHECK-P8-NEXT: mtvsrd v0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI6_1@toc@ha
; CHECK-P8-NEXT: addi r5, r5, .LCPI6_0@toc@l
; CHECK-P8-NEXT: addi r4, r4, .LCPI6_1@toc@l
; CHECK-P8-NEXT: lxvd2x vs0, 0, r5
; CHECK-P8-NEXT: addis r5, r2, .LCPI6_2@toc@ha
; CHECK-P8-NEXT: addi r5, r5, .LCPI6_2@toc@l
; CHECK-P8-NEXT: lxvd2x vs1, 0, r5
; CHECK-P8-NEXT: addis r5, r2, .LCPI6_4@toc@ha
; CHECK-P8-NEXT: addi r5, r5, .LCPI6_4@toc@l
; CHECK-P8-NEXT: lxvd2x vs2, 0, r5
; CHECK-P8-NEXT: addis r5, r2, .LCPI6_3@toc@ha
; CHECK-P8-NEXT: addi r5, r5, .LCPI6_3@toc@l
; CHECK-P8-NEXT: xxswapd v3, vs0
; CHECK-P8-NEXT: vperm v3, v0, v0, v3
; CHECK-P8-NEXT: xxswapd v4, vs1
; CHECK-P8-NEXT: vperm v4, v0, v0, v4
; CHECK-P8-NEXT: xxswapd v2, vs2
; CHECK-P8-NEXT: lxvd2x vs2, 0, r5
; CHECK-P8-NEXT: vperm v2, v0, v0, v2
; CHECK-P8-NEXT: xxswapd v5, vs2
; CHECK-P8-NEXT: vperm v5, v0, v0, v5
; CHECK-P8-NEXT: lxvd2x v0, 0, r4
; CHECK-P8-NEXT: li r4, 48
; CHECK-P8-NEXT: vsld v2, v2, v0
; CHECK-P8-NEXT: vsld v5, v5, v0
; CHECK-P8-NEXT: vsld v3, v3, v0
; CHECK-P8-NEXT: vsld v4, v4, v0
; CHECK-P8-NEXT: vsrad v2, v2, v0
; CHECK-P8-NEXT: vsrad v5, v5, v0
; CHECK-P8-NEXT: vsrad v3, v3, v0
; CHECK-P8-NEXT: vsrad v4, v4, v0
; CHECK-P8-NEXT: xvcvsxddp vs2, v2
; CHECK-P8-NEXT: xvcvsxddp vs3, v5
; CHECK-P8-NEXT: xvcvsxddp vs0, v3
; CHECK-P8-NEXT: xvcvsxddp vs1, v4
; CHECK-P8-NEXT: xxswapd vs2, vs2
; CHECK-P8-NEXT: xxswapd vs3, vs3
; CHECK-P8-NEXT: xxswapd vs1, vs1
; CHECK-P8-NEXT: xxswapd vs0, vs0
; CHECK-P8-NEXT: stxvd2x vs2, r3, r4
; CHECK-P8-NEXT: li r4, 32
; CHECK-P8-NEXT: stxvd2x vs3, r3, r4
; CHECK-P8-NEXT: li r4, 16
; CHECK-P8-NEXT: stxvd2x vs1, r3, r4
; CHECK-P8-NEXT: stxvd2x vs0, 0, r3
; CHECK-P8-NEXT: blr
;
; CHECK-P9-LABEL: test8elt_signed:
; CHECK-P9: # %bb.0: # %entry
; CHECK-P9-NEXT: mtvsrd v2, r4
; CHECK-P9-NEXT: addis r4, r2, .LCPI6_0@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI6_0@toc@l
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI6_1@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI6_1@toc@l
; CHECK-P9-NEXT: vperm v3, v2, v2, v3
; CHECK-P9-NEXT: vextsb2d v3, v3
; CHECK-P9-NEXT: xvcvsxddp vs0, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI6_2@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI6_2@toc@l
; CHECK-P9-NEXT: vperm v3, v2, v2, v3
; CHECK-P9-NEXT: stxv vs0, 0(r3)
; CHECK-P9-NEXT: vextsb2d v3, v3
; CHECK-P9-NEXT: xvcvsxddp vs1, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI6_3@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI6_3@toc@l
; CHECK-P9-NEXT: vperm v3, v2, v2, v3
; CHECK-P9-NEXT: stxv vs1, 16(r3)
; CHECK-P9-NEXT: vextsb2d v3, v3
; CHECK-P9-NEXT: xvcvsxddp vs2, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: vperm v2, v2, v2, v3
; CHECK-P9-NEXT: stxv vs2, 32(r3)
; CHECK-P9-NEXT: vextsb2d v2, v2
; CHECK-P9-NEXT: xvcvsxddp vs3, v2
; CHECK-P9-NEXT: stxv vs3, 48(r3)
; CHECK-P9-NEXT: blr
;
; CHECK-BE-LABEL: test8elt_signed:
; CHECK-BE: # %bb.0: # %entry
; CHECK-BE-NEXT: mtvsrd v2, r4
; CHECK-BE-NEXT: addis r4, r2, .LCPI6_0@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI6_0@toc@l
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI6_1@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI6_1@toc@l
; CHECK-BE-NEXT: vperm v3, v2, v2, v3
; CHECK-BE-NEXT: vextsb2d v3, v3
; CHECK-BE-NEXT: xvcvsxddp vs0, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI6_2@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI6_2@toc@l
; CHECK-BE-NEXT: vperm v3, v2, v2, v3
; CHECK-BE-NEXT: stxv vs0, 0(r3)
; CHECK-BE-NEXT: vextsb2d v3, v3
; CHECK-BE-NEXT: xvcvsxddp vs1, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI6_3@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI6_3@toc@l
; CHECK-BE-NEXT: vperm v3, v2, v2, v3
; CHECK-BE-NEXT: stxv vs1, 16(r3)
; CHECK-BE-NEXT: vextsb2d v3, v3
; CHECK-BE-NEXT: xvcvsxddp vs2, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: vperm v2, v2, v2, v3
; CHECK-BE-NEXT: stxv vs2, 32(r3)
; CHECK-BE-NEXT: vextsb2d v2, v2
; CHECK-BE-NEXT: xvcvsxddp vs3, v2
; CHECK-BE-NEXT: stxv vs3, 48(r3)
; CHECK-BE-NEXT: blr
entry:
%0 = bitcast i64 %a.coerce to <8 x i8>
%1 = sitofp <8 x i8> %0 to <8 x double>
store <8 x double> %1, ptr %agg.result, align 64
ret void
}
define void @test16elt_signed(ptr noalias nocapture sret(<16 x double>) %agg.result, <16 x i8> %a) local_unnamed_addr #2 {
; CHECK-P8-LABEL: test16elt_signed:
; CHECK-P8: # %bb.0: # %entry
; CHECK-P8-NEXT: addis r4, r2, .LCPI7_0@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI7_0@toc@l
; CHECK-P8-NEXT: lxvd2x vs0, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI7_2@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI7_2@toc@l
; CHECK-P8-NEXT: lxvd2x vs1, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI7_4@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI7_4@toc@l
; CHECK-P8-NEXT: lxvd2x vs2, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI7_7@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI7_7@toc@l
; CHECK-P8-NEXT: lxvd2x vs3, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI7_8@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI7_8@toc@l
; CHECK-P8-NEXT: lxvd2x vs4, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI7_6@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI7_6@toc@l
; CHECK-P8-NEXT: xxswapd v6, vs0
; CHECK-P8-NEXT: vperm v6, v2, v2, v6
; CHECK-P8-NEXT: xxswapd v7, vs1
; CHECK-P8-NEXT: vperm v7, v2, v2, v7
; CHECK-P8-NEXT: xxswapd v0, vs2
; CHECK-P8-NEXT: vperm v0, v2, v2, v0
; CHECK-P8-NEXT: xxswapd v3, vs3
; CHECK-P8-NEXT: lxvd2x vs3, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI7_5@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI7_5@toc@l
; CHECK-P8-NEXT: vperm v3, v2, v2, v3
; CHECK-P8-NEXT: xxswapd v4, vs4
; CHECK-P8-NEXT: vperm v4, v2, v2, v4
; CHECK-P8-NEXT: xxswapd v5, vs3
; CHECK-P8-NEXT: lxvd2x vs3, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI7_3@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI7_3@toc@l
; CHECK-P8-NEXT: lxvd2x vs2, 0, r4
; CHECK-P8-NEXT: addis r4, r2, .LCPI7_1@toc@ha
; CHECK-P8-NEXT: addi r4, r4, .LCPI7_1@toc@l
; CHECK-P8-NEXT: vperm v5, v2, v2, v5
; CHECK-P8-NEXT: xxswapd v1, vs3
; CHECK-P8-NEXT: vperm v1, v2, v2, v1
; CHECK-P8-NEXT: xxswapd v8, vs2
; CHECK-P8-NEXT: vperm v2, v2, v2, v8
; CHECK-P8-NEXT: lxvd2x v8, 0, r4
; CHECK-P8-NEXT: li r4, 112
; CHECK-P8-NEXT: vsld v4, v4, v8
; CHECK-P8-NEXT: vsld v3, v3, v8
; CHECK-P8-NEXT: vsld v5, v5, v8
; CHECK-P8-NEXT: vsld v1, v1, v8
; CHECK-P8-NEXT: vsld v0, v0, v8
; CHECK-P8-NEXT: vsld v2, v2, v8
; CHECK-P8-NEXT: vsld v6, v6, v8
; CHECK-P8-NEXT: vsld v7, v7, v8
; CHECK-P8-NEXT: vsrad v4, v4, v8
; CHECK-P8-NEXT: vsrad v3, v3, v8
; CHECK-P8-NEXT: vsrad v5, v5, v8
; CHECK-P8-NEXT: vsrad v1, v1, v8
; CHECK-P8-NEXT: vsrad v0, v0, v8
; CHECK-P8-NEXT: vsrad v2, v2, v8
; CHECK-P8-NEXT: vsrad v6, v6, v8
; CHECK-P8-NEXT: vsrad v7, v7, v8
; CHECK-P8-NEXT: xvcvsxddp vs4, v4
; CHECK-P8-NEXT: xvcvsxddp vs3, v3
; CHECK-P8-NEXT: xvcvsxddp vs7, v5
; CHECK-P8-NEXT: xvcvsxddp vs6, v1
; CHECK-P8-NEXT: xvcvsxddp vs5, v0
; CHECK-P8-NEXT: xvcvsxddp vs2, v2
; CHECK-P8-NEXT: xvcvsxddp vs0, v6
; CHECK-P8-NEXT: xvcvsxddp vs1, v7
; CHECK-P8-NEXT: xxswapd vs4, vs4
; CHECK-P8-NEXT: xxswapd vs3, vs3
; CHECK-P8-NEXT: xxswapd vs7, vs7
; CHECK-P8-NEXT: xxswapd vs2, vs2
; CHECK-P8-NEXT: xxswapd vs1, vs1
; CHECK-P8-NEXT: xxswapd vs0, vs0
; CHECK-P8-NEXT: stxvd2x vs4, r3, r4
; CHECK-P8-NEXT: li r4, 96
; CHECK-P8-NEXT: xxswapd vs4, vs5
; CHECK-P8-NEXT: stxvd2x vs3, r3, r4
; CHECK-P8-NEXT: li r4, 80
; CHECK-P8-NEXT: stxvd2x vs7, r3, r4
; CHECK-P8-NEXT: li r4, 64
; CHECK-P8-NEXT: xxswapd vs3, vs6
; CHECK-P8-NEXT: stxvd2x vs3, r3, r4
; CHECK-P8-NEXT: li r4, 48
; CHECK-P8-NEXT: stxvd2x vs4, r3, r4
; CHECK-P8-NEXT: li r4, 32
; CHECK-P8-NEXT: stxvd2x vs2, r3, r4
; CHECK-P8-NEXT: li r4, 16
; CHECK-P8-NEXT: stxvd2x vs1, r3, r4
; CHECK-P8-NEXT: stxvd2x vs0, 0, r3
; CHECK-P8-NEXT: blr
;
; CHECK-P9-LABEL: test16elt_signed:
; CHECK-P9: # %bb.0: # %entry
; CHECK-P9-NEXT: addis r4, r2, .LCPI7_0@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI7_0@toc@l
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI7_1@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI7_1@toc@l
; CHECK-P9-NEXT: vperm v3, v2, v2, v3
; CHECK-P9-NEXT: vextsb2d v3, v3
; CHECK-P9-NEXT: xvcvsxddp vs0, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI7_2@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI7_2@toc@l
; CHECK-P9-NEXT: vperm v3, v2, v2, v3
; CHECK-P9-NEXT: stxv vs0, 0(r3)
; CHECK-P9-NEXT: vextsb2d v3, v3
; CHECK-P9-NEXT: xvcvsxddp vs1, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI7_3@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI7_3@toc@l
; CHECK-P9-NEXT: vperm v3, v2, v2, v3
; CHECK-P9-NEXT: stxv vs1, 16(r3)
; CHECK-P9-NEXT: vextsb2d v3, v3
; CHECK-P9-NEXT: xvcvsxddp vs2, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI7_4@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI7_4@toc@l
; CHECK-P9-NEXT: vperm v3, v2, v2, v3
; CHECK-P9-NEXT: stxv vs2, 32(r3)
; CHECK-P9-NEXT: vextsb2d v3, v3
; CHECK-P9-NEXT: xvcvsxddp vs3, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI7_5@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI7_5@toc@l
; CHECK-P9-NEXT: vperm v3, v2, v2, v3
; CHECK-P9-NEXT: stxv vs3, 48(r3)
; CHECK-P9-NEXT: vextsb2d v3, v3
; CHECK-P9-NEXT: xvcvsxddp vs4, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI7_6@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI7_6@toc@l
; CHECK-P9-NEXT: vperm v3, v2, v2, v3
; CHECK-P9-NEXT: stxv vs4, 64(r3)
; CHECK-P9-NEXT: vextsb2d v3, v3
; CHECK-P9-NEXT: xvcvsxddp vs5, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: addis r4, r2, .LCPI7_7@toc@ha
; CHECK-P9-NEXT: addi r4, r4, .LCPI7_7@toc@l
; CHECK-P9-NEXT: vperm v3, v2, v2, v3
; CHECK-P9-NEXT: stxv vs5, 80(r3)
; CHECK-P9-NEXT: vextsb2d v3, v3
; CHECK-P9-NEXT: xvcvsxddp vs6, v3
; CHECK-P9-NEXT: lxv v3, 0(r4)
; CHECK-P9-NEXT: vperm v2, v2, v2, v3
; CHECK-P9-NEXT: stxv vs6, 96(r3)
; CHECK-P9-NEXT: vextsb2d v2, v2
; CHECK-P9-NEXT: xvcvsxddp vs7, v2
; CHECK-P9-NEXT: stxv vs7, 112(r3)
; CHECK-P9-NEXT: blr
;
; CHECK-BE-LABEL: test16elt_signed:
; CHECK-BE: # %bb.0: # %entry
; CHECK-BE-NEXT: addis r4, r2, .LCPI7_0@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI7_0@toc@l
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI7_1@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI7_1@toc@l
; CHECK-BE-NEXT: vperm v3, v2, v2, v3
; CHECK-BE-NEXT: vextsb2d v3, v3
; CHECK-BE-NEXT: xvcvsxddp vs0, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI7_2@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI7_2@toc@l
; CHECK-BE-NEXT: vperm v3, v2, v2, v3
; CHECK-BE-NEXT: stxv vs0, 0(r3)
; CHECK-BE-NEXT: vextsb2d v3, v3
; CHECK-BE-NEXT: xvcvsxddp vs1, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI7_3@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI7_3@toc@l
; CHECK-BE-NEXT: vperm v3, v2, v2, v3
; CHECK-BE-NEXT: stxv vs1, 16(r3)
; CHECK-BE-NEXT: vextsb2d v3, v3
; CHECK-BE-NEXT: xvcvsxddp vs2, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI7_4@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI7_4@toc@l
; CHECK-BE-NEXT: vperm v3, v2, v2, v3
; CHECK-BE-NEXT: stxv vs2, 32(r3)
; CHECK-BE-NEXT: vextsb2d v3, v3
; CHECK-BE-NEXT: xvcvsxddp vs3, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI7_5@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI7_5@toc@l
; CHECK-BE-NEXT: vperm v3, v2, v2, v3
; CHECK-BE-NEXT: stxv vs3, 48(r3)
; CHECK-BE-NEXT: vextsb2d v3, v3
; CHECK-BE-NEXT: xvcvsxddp vs4, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI7_6@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI7_6@toc@l
; CHECK-BE-NEXT: vperm v3, v2, v2, v3
; CHECK-BE-NEXT: stxv vs4, 64(r3)
; CHECK-BE-NEXT: vextsb2d v3, v3
; CHECK-BE-NEXT: xvcvsxddp vs5, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: addis r4, r2, .LCPI7_7@toc@ha
; CHECK-BE-NEXT: addi r4, r4, .LCPI7_7@toc@l
; CHECK-BE-NEXT: vperm v3, v2, v2, v3
; CHECK-BE-NEXT: stxv vs5, 80(r3)
; CHECK-BE-NEXT: vextsb2d v3, v3
; CHECK-BE-NEXT: xvcvsxddp vs6, v3
; CHECK-BE-NEXT: lxv v3, 0(r4)
; CHECK-BE-NEXT: vperm v2, v2, v2, v3
; CHECK-BE-NEXT: stxv vs6, 96(r3)
; CHECK-BE-NEXT: vextsb2d v2, v2
; CHECK-BE-NEXT: xvcvsxddp vs7, v2
; CHECK-BE-NEXT: stxv vs7, 112(r3)
; CHECK-BE-NEXT: blr
entry:
%0 = sitofp <16 x i8> %a to <16 x double>
store <16 x double> %0, ptr %agg.result, align 128
ret void
}
Codebase Browser
llvm