summaryrefslogtreecommitdiffstats
path: root/llvm/test/CodeGen/X86/stack-folding-int-avx2.ll
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/test/CodeGen/X86/stack-folding-int-avx2.ll')
-rw-r--r--llvm/test/CodeGen/X86/stack-folding-int-avx2.ll45
1 files changed, 24 insertions, 21 deletions
diff --git a/llvm/test/CodeGen/X86/stack-folding-int-avx2.ll b/llvm/test/CodeGen/X86/stack-folding-int-avx2.ll
index ef7fa221714..897c749cc0e 100644
--- a/llvm/test/CodeGen/X86/stack-folding-int-avx2.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-int-avx2.ll
@@ -662,19 +662,19 @@ define <8 x i32> @stack_fold_pmovsxbd(<16 x i8> %a0) {
;CHECK-LABEL: stack_fold_pmovsxbd
;CHECK: vpmovsxbd {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
- %2 = shufflevector <16 x i8> %a0, <16 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
- %3 = sext <8 x i8> %2 to <8 x i32>
- ret <8 x i32> %3
+ %2 = call <8 x i32> @llvm.x86.avx2.pmovsxbd(<16 x i8> %a0)
+ ret <8 x i32> %2
}
+declare <8 x i32> @llvm.x86.avx2.pmovsxbd(<16 x i8>) nounwind readnone
define <4 x i64> @stack_fold_pmovsxbq(<16 x i8> %a0) {
;CHECK-LABEL: stack_fold_pmovsxbq
;CHECK: pmovsxbq {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
- %2 = shufflevector <16 x i8> %a0, <16 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
- %3 = sext <4 x i8> %2 to <4 x i64>
- ret <4 x i64> %3
+ %2 = call <4 x i64> @llvm.x86.avx2.pmovsxbq(<16 x i8> %a0)
+ ret <4 x i64> %2
}
+declare <4 x i64> @llvm.x86.avx2.pmovsxbq(<16 x i8>) nounwind readnone
define <16 x i16> @stack_fold_pmovsxbw(<16 x i8> %a0) {
;CHECK-LABEL: stack_fold_pmovsxbw
@@ -704,61 +704,64 @@ define <4 x i64> @stack_fold_pmovsxwq(<8 x i16> %a0) {
;CHECK-LABEL: stack_fold_pmovsxwq
;CHECK: vpmovsxwq {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
- %2 = shufflevector <8 x i16> %a0, <8 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
- %3 = sext <4 x i16> %2 to <4 x i64>
- ret <4 x i64> %3
+ %2 = call <4 x i64> @llvm.x86.avx2.pmovsxwq(<8 x i16> %a0)
+ ret <4 x i64> %2
}
+declare <4 x i64> @llvm.x86.avx2.pmovsxwq(<8 x i16>) nounwind readnone
define <8 x i32> @stack_fold_pmovzxbd(<16 x i8> %a0) {
;CHECK-LABEL: stack_fold_pmovzxbd
;CHECK: vpmovzxbd {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
- %2 = shufflevector <16 x i8> %a0, <16 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
- %3 = zext <8 x i8> %2 to <8 x i32>
- ret <8 x i32> %3
+ %2 = call <8 x i32> @llvm.x86.avx2.pmovzxbd(<16 x i8> %a0)
+ ret <8 x i32> %2
}
+declare <8 x i32> @llvm.x86.avx2.pmovzxbd(<16 x i8>) nounwind readnone
define <4 x i64> @stack_fold_pmovzxbq(<16 x i8> %a0) {
;CHECK-LABEL: stack_fold_pmovzxbq
;CHECK: vpmovzxbq {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
- %2 = shufflevector <16 x i8> %a0, <16 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
- %3 = zext <4 x i8> %2 to <4 x i64>
- ret <4 x i64> %3
+ %2 = call <4 x i64> @llvm.x86.avx2.pmovzxbq(<16 x i8> %a0)
+ ret <4 x i64> %2
}
+declare <4 x i64> @llvm.x86.avx2.pmovzxbq(<16 x i8>) nounwind readnone
define <16 x i16> @stack_fold_pmovzxbw(<16 x i8> %a0) {
;CHECK-LABEL: stack_fold_pmovzxbw
;CHECK: vpmovzxbw {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
- %2 = zext <16 x i8> %a0 to <16 x i16>
+ %2 = call <16 x i16> @llvm.x86.avx2.pmovzxbw(<16 x i8> %a0)
ret <16 x i16> %2
}
+declare <16 x i16> @llvm.x86.avx2.pmovzxbw(<16 x i8>) nounwind readnone
define <4 x i64> @stack_fold_pmovzxdq(<4 x i32> %a0) {
;CHECK-LABEL: stack_fold_pmovzxdq
;CHECK: vpmovzxdq {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
- %2 = zext <4 x i32> %a0 to <4 x i64>
+ %2 = call <4 x i64> @llvm.x86.avx2.pmovzxdq(<4 x i32> %a0)
ret <4 x i64> %2
}
+declare <4 x i64> @llvm.x86.avx2.pmovzxdq(<4 x i32>) nounwind readnone
define <8 x i32> @stack_fold_pmovzxwd(<8 x i16> %a0) {
;CHECK-LABEL: stack_fold_pmovzxwd
;CHECK: vpmovzxwd {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
- %2 = zext <8 x i16> %a0 to <8 x i32>
+ %2 = call <8 x i32> @llvm.x86.avx2.pmovzxwd(<8 x i16> %a0)
ret <8 x i32> %2
}
+declare <8 x i32> @llvm.x86.avx2.pmovzxwd(<8 x i16>) nounwind readnone
define <4 x i64> @stack_fold_pmovzxwq(<8 x i16> %a0) {
;CHECK-LABEL: stack_fold_pmovzxwq
;CHECK: vpmovzxwq {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
%1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
- %2 = shufflevector <8 x i16> %a0, <8 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
- %3 = zext <4 x i16> %2 to <4 x i64>
- ret <4 x i64> %3
+ %2 = call <4 x i64> @llvm.x86.avx2.pmovzxwq(<8 x i16> %a0)
+ ret <4 x i64> %2
}
+declare <4 x i64> @llvm.x86.avx2.pmovzxwq(<8 x i16>) nounwind readnone
define <4 x i64> @stack_fold_pmuldq(<8 x i32> %a0, <8 x i32> %a1) {
;CHECK-LABEL: stack_fold_pmuldq
OpenPOWER on IntegriCloud