diff options
Diffstat (limited to 'test')
| -rw-r--r-- | test/Assembler/AutoUpgradeIntrinsics.ll | 18 | ||||
| -rw-r--r-- | test/CodeGen/X86/nontemporal.ll | 19 |
2 files changed, 37 insertions, 0 deletions
diff --git a/test/Assembler/AutoUpgradeIntrinsics.ll b/test/Assembler/AutoUpgradeIntrinsics.ll index e4e2d3a56e..417493f716 100644 --- a/test/Assembler/AutoUpgradeIntrinsics.ll +++ b/test/Assembler/AutoUpgradeIntrinsics.ll @@ -10,6 +10,7 @@ ; RUN: not grep {llvm\\.x86\\.sse2\\.loadu} ; RUN: llvm-as < %s | llvm-dis | \ ; RUN: grep {llvm\\.x86\\.mmx\\.ps} | grep {x86_mmx} | count 16 +; RUN: llvm-as < %s | llvm-dis | FileCheck %s declare i32 @llvm.ctpop.i28(i28 %val) declare i32 @llvm.cttz.i29(i29 %val) @@ -91,3 +92,20 @@ define void @test_loadu(i8* %a, double* %b) { %v2 = call <2 x double> @llvm.x86.sse2.loadu.pd(double* %b) ret void } + +declare void @llvm.x86.sse.movnt.ps(i8*, <4 x float>) nounwind readnone +declare void @llvm.x86.sse2.movnt.dq(i8*, <2 x double>) nounwind readnone +declare void @llvm.x86.sse2.movnt.pd(i8*, <2 x double>) nounwind readnone +declare void @llvm.x86.sse2.movnt.i(i8*, i32) nounwind readnone + +define void @f(<4 x float> %A, i8* %B, <2 x double> %C, i32 %D) { +; CHECK: store{{.*}}nontemporal + call void @llvm.x86.sse.movnt.ps(i8* %B, <4 x float> %A) +; CHECK: store{{.*}}nontemporal + call void @llvm.x86.sse2.movnt.dq(i8* %B, <2 x double> %C) +; CHECK: store{{.*}}nontemporal + call void @llvm.x86.sse2.movnt.pd(i8* %B, <2 x double> %C) +; CHECK: store{{.*}}nontemporal + call void @llvm.x86.sse2.movnt.i(i8* %B, i32 %D) + ret void +} diff --git a/test/CodeGen/X86/nontemporal.ll b/test/CodeGen/X86/nontemporal.ll new file mode 100644 index 0000000000..1d095359b6 --- /dev/null +++ b/test/CodeGen/X86/nontemporal.ll @@ -0,0 +1,19 @@ +; RUN: llc < %s -march=x86 -mattr=+sse2 | FileCheck %s + +define void @f(<4 x float> %A, i8* %B, <2 x double> %C, i32 %D, <2 x i64> %E) { +; CHECK: movntps + %cast = bitcast i8* %B to <4 x float>* + store <4 x float> %A, <4 x float>* %cast, align 16, !nontemporal !0 +; CHECK: movntdq + %cast1 = bitcast i8* %B to <2 x i64>* + store <2 x i64> %E, <2 x i64>* %cast1, align 16, !nontemporal !0 +; CHECK: movntpd + %cast2 = bitcast i8* %B to <2 x double>* + store <2 x double> %C, <2 x double>* %cast2, align 16, !nontemporal !0 +; CHECK: movnti + %cast3 = bitcast i8* %B to i32* + store i32 %D, i32* %cast3, align 16, !nontemporal !0 + ret void +} + +!0 = metadata !{i32 1} |