summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--llvm/test/CodeGen/X86/stack-folding-fp-avx1.ll33
-rw-r--r--llvm/test/CodeGen/X86/stack-folding-fp-sse42.ll18
2 files changed, 51 insertions, 0 deletions
diff --git a/llvm/test/CodeGen/X86/stack-folding-fp-avx1.ll b/llvm/test/CodeGen/X86/stack-folding-fp-avx1.ll
index b86ec0ea22f..bb504fa29f9 100644
--- a/llvm/test/CodeGen/X86/stack-folding-fp-avx1.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-fp-avx1.ll
@@ -384,6 +384,14 @@ define <2 x double> @stack_fold_cvtdq2pd(<4 x i32> %a0) {
;CHECK-LABEL: stack_fold_cvtdq2pd
;CHECK: vcvtdq2pd {{-?[0-9]*}}(%rsp), {{%xmm[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 <4 x i32> %a0, <4 x i32> undef, <2 x i32> <i32 0, i32 1>
+ %3 = sitofp <2 x i32> %2 to <2 x double>
+ ret <2 x double> %3
+}
+define <2 x double> @stack_fold_cvtdq2pd_int(<4 x i32> %a0) {
+ ;CHECK-LABEL: stack_fold_cvtdq2pd_int
+ ;CHECK: vcvtdq2pd {{-?[0-9]*}}(%rsp), {{%xmm[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 = call <2 x double> @llvm.x86.sse2.cvtdq2pd(<4 x i32> %a0)
ret <2 x double> %2
}
@@ -393,6 +401,14 @@ define <4 x double> @stack_fold_cvtdq2pd_ymm(<4 x i32> %a0) {
;CHECK-LABEL: stack_fold_cvtdq2pd_ymm
;CHECK: vcvtdq2pd {{-?[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 = sitofp <4 x i32> %a0 to <4 x double>
+ ret <4 x double> %2
+}
+
+define <4 x double> @stack_fold_cvtdq2pd_ymm_int(<4 x i32> %a0) {
+ ;CHECK-LABEL: stack_fold_cvtdq2pd_ymm_int
+ ;CHECK: vcvtdq2pd {{-?[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 = call <4 x double> @llvm.x86.avx.cvtdq2.pd.256(<4 x i32> %a0)
ret <4 x double> %2
}
@@ -488,6 +504,15 @@ define <2 x double> @stack_fold_cvtps2pd(<4 x float> %a0) {
;CHECK-LABEL: stack_fold_cvtps2pd
;CHECK: vcvtps2pd {{-?[0-9]*}}(%rsp), {{%xmm[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 <4 x float> %a0, <4 x float> undef, <2 x i32> <i32 0, i32 1>
+ %3 = fpext <2 x float> %2 to <2 x double>
+ ret <2 x double> %3
+}
+
+define <2 x double> @stack_fold_cvtps2pd_int(<4 x float> %a0) {
+ ;CHECK-LABEL: stack_fold_cvtps2pd_int
+ ;CHECK: vcvtps2pd {{-?[0-9]*}}(%rsp), {{%xmm[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 = call <2 x double> @llvm.x86.sse2.cvtps2pd(<4 x float> %a0)
ret <2 x double> %2
}
@@ -497,6 +522,14 @@ define <4 x double> @stack_fold_cvtps2pd_ymm(<4 x float> %a0) {
;CHECK-LABEL: stack_fold_cvtps2pd_ymm
;CHECK: vcvtps2pd {{-?[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 = fpext <4 x float> %a0 to <4 x double>
+ ret <4 x double> %2
+}
+
+define <4 x double> @stack_fold_cvtps2pd_ymm_int(<4 x float> %a0) {
+ ;CHECK-LABEL: stack_fold_cvtps2pd_ymm_int
+ ;CHECK: vcvtps2pd {{-?[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 = call <4 x double> @llvm.x86.avx.cvt.ps2.pd.256(<4 x float> %a0)
ret <4 x double> %2
}
diff --git a/llvm/test/CodeGen/X86/stack-folding-fp-sse42.ll b/llvm/test/CodeGen/X86/stack-folding-fp-sse42.ll
index 9f689cfe85e..4c675356df6 100644
--- a/llvm/test/CodeGen/X86/stack-folding-fp-sse42.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-fp-sse42.ll
@@ -244,6 +244,15 @@ define <2 x double> @stack_fold_cvtdq2pd(<4 x i32> %a0) {
;CHECK-LABEL: stack_fold_cvtdq2pd
;CHECK: cvtdq2pd {{-?[0-9]*}}(%rsp), {{%xmm[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 <4 x i32> %a0, <4 x i32> undef, <2 x i32> <i32 0, i32 1>
+ %3 = sitofp <2 x i32> %2 to <2 x double>
+ ret <2 x double> %3
+}
+
+define <2 x double> @stack_fold_cvtdq2pd_int(<4 x i32> %a0) {
+ ;CHECK-LABEL: stack_fold_cvtdq2pd_int
+ ;CHECK: cvtdq2pd {{-?[0-9]*}}(%rsp), {{%xmm[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 = call <2 x double> @llvm.x86.sse2.cvtdq2pd(<4 x i32> %a0)
ret <2 x double> %2
}
@@ -287,6 +296,15 @@ define <2 x double> @stack_fold_cvtps2pd(<4 x float> %a0) {
;CHECK-LABEL: stack_fold_cvtps2pd
;CHECK: cvtps2pd {{-?[0-9]*}}(%rsp), {{%xmm[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 <4 x float> %a0, <4 x float> undef, <2 x i32> <i32 0, i32 1>
+ %3 = fpext <2 x float> %2 to <2 x double>
+ ret <2 x double> %3
+}
+
+define <2 x double> @stack_fold_cvtps2pd_int(<4 x float> %a0) {
+ ;CHECK-LABEL: stack_fold_cvtps2pd_int
+ ;CHECK: cvtps2pd {{-?[0-9]*}}(%rsp), {{%xmm[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 = call <2 x double> @llvm.x86.sse2.cvtps2pd(<4 x float> %a0)
ret <2 x double> %2
}
OpenPOWER on IntegriCloud