1 // RUN: %clang_cc1 %s -triple=x86_64-apple-darwin -target-feature +sse3 -emit-llvm -o - -Werror | FileCheck %s
2
3 // Don't include mm_malloc.h, it's system specific.
4 #define __MM_MALLOC_H
5
6 #include <x86intrin.h>
7
8 // NOTE: This should match the tests in llvm/test/CodeGen/X86/sse3-intrinsics-fast-isel.ll
9
test_mm_addsub_pd(__m128d A,__m128d B)10 __m128d test_mm_addsub_pd(__m128d A, __m128d B) {
11 // CHECK-LABEL: test_mm_addsub_pd
12 // CHECK: call <2 x double> @llvm.x86.sse3.addsub.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
13 return _mm_addsub_pd(A, B);
14 }
15
test_mm_addsub_ps(__m128 A,__m128 B)16 __m128 test_mm_addsub_ps(__m128 A, __m128 B) {
17 // CHECK-LABEL: test_mm_addsub_ps
18 // CHECK: call <4 x float> @llvm.x86.sse3.addsub.ps(<4 x float> %{{.*}}, <4 x float> %{{.*}})
19 return _mm_addsub_ps(A, B);
20 }
21
test_mm_hadd_pd(__m128d A,__m128d B)22 __m128d test_mm_hadd_pd(__m128d A, __m128d B) {
23 // CHECK-LABEL: test_mm_hadd_pd
24 // CHECK: call <2 x double> @llvm.x86.sse3.hadd.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
25 return _mm_hadd_pd(A, B);
26 }
27
test_mm_hadd_ps(__m128 A,__m128 B)28 __m128 test_mm_hadd_ps(__m128 A, __m128 B) {
29 // CHECK-LABEL: test_mm_hadd_ps
30 // CHECK: call <4 x float> @llvm.x86.sse3.hadd.ps(<4 x float> %{{.*}}, <4 x float> %{{.*}})
31 return _mm_hadd_ps(A, B);
32 }
33
test_mm_hsub_pd(__m128d A,__m128d B)34 __m128d test_mm_hsub_pd(__m128d A, __m128d B) {
35 // CHECK-LABEL: test_mm_hsub_pd
36 // CHECK: call <2 x double> @llvm.x86.sse3.hsub.pd(<2 x double> %{{.*}}, <2 x double> %{{.*}})
37 return _mm_hsub_pd(A, B);
38 }
39
test_mm_hsub_ps(__m128 A,__m128 B)40 __m128 test_mm_hsub_ps(__m128 A, __m128 B) {
41 // CHECK-LABEL: test_mm_hsub_ps
42 // CHECK: call <4 x float> @llvm.x86.sse3.hsub.ps(<4 x float> %{{.*}}, <4 x float> %{{.*}})
43 return _mm_hsub_ps(A, B);
44 }
45
test_mm_lddqu_si128(__m128i const * P)46 __m128i test_mm_lddqu_si128(__m128i const* P) {
47 // CHECK-LABEL: test_mm_lddqu_si128
48 // CHECK: call <16 x i8> @llvm.x86.sse3.ldu.dq(i8* %{{.*}})
49 return _mm_lddqu_si128(P);
50 }
51
test_mm_loaddup_pd(double const * P)52 __m128d test_mm_loaddup_pd(double const* P) {
53 // CHECK-LABEL: test_mm_loaddup_pd
54 // CHECK: load double*
55 // CHECK: insertelement <2 x double> undef, double %{{.*}}, i32 0
56 // CHECK: insertelement <2 x double> %{{.*}}, double %{{.*}}, i32 1
57 return _mm_loaddup_pd(P);
58 }
59
test_mm_movedup_pd(__m128d A)60 __m128d test_mm_movedup_pd(__m128d A) {
61 // CHECK-LABEL: test_mm_movedup_pd
62 // CHECK: shufflevector <2 x double> %{{.*}}, <2 x double> %{{.*}}, <2 x i32> zeroinitializer
63 return _mm_movedup_pd(A);
64 }
65
test_mm_movehdup_ps(__m128 A)66 __m128 test_mm_movehdup_ps(__m128 A) {
67 // CHECK-LABEL: test_mm_movehdup_ps
68 // CHECK: shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <4 x i32> <i32 1, i32 1, i32 3, i32 3>
69 return _mm_movehdup_ps(A);
70 }
71
test_mm_moveldup_ps(__m128 A)72 __m128 test_mm_moveldup_ps(__m128 A) {
73 // CHECK-LABEL: test_mm_moveldup_ps
74 // CHECK: shufflevector <4 x float> %{{.*}}, <4 x float> %{{.*}}, <4 x i32> <i32 0, i32 0, i32 2, i32 2>
75 return _mm_moveldup_ps(A);
76 }
77