1 // RUN: %clang_cc1 %s -triple=x86_64-apple-darwin -target-feature +f16c -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 float test_cvtsh_ss(unsigned short a) { 9 // CHECK-LABEL: test_cvtsh_ss 10 // CHECK: insertelement <8 x i16> undef, i16 %{{.*}}, i32 0 11 // CHECK: insertelement <8 x i16> %{{.*}}, i16 0, i32 1 12 // CHECK: insertelement <8 x i16> %{{.*}}, i16 0, i32 2 13 // CHECK: insertelement <8 x i16> %{{.*}}, i16 0, i32 3 14 // CHECK: insertelement <8 x i16> %{{.*}}, i16 0, i32 4 15 // CHECK: insertelement <8 x i16> %{{.*}}, i16 0, i32 5 16 // CHECK: insertelement <8 x i16> %{{.*}}, i16 0, i32 6 17 // CHECK: insertelement <8 x i16> %{{.*}}, i16 0, i32 7 18 // CHECK: call <4 x float> @llvm.x86.vcvtph2ps.128(<8 x i16> %{{.*}}) 19 // CHECK: extractelement <4 x float> %{{.*}}, i32 0 20 return _cvtsh_ss(a); 21 } 22 23 unsigned short test_cvtss_sh(float a) { 24 // CHECK-LABEL: test_cvtss_sh 25 // CHECK: insertelement <4 x float> undef, float %{{.*}}, i32 0 26 // CHECK: insertelement <4 x float> %{{.*}}, float 0.000000e+00, i32 1 27 // CHECK: insertelement <4 x float> %{{.*}}, float 0.000000e+00, i32 2 28 // CHECK: insertelement <4 x float> %{{.*}}, float 0.000000e+00, i32 3 29 // CHECK: call <8 x i16> @llvm.x86.vcvtps2ph.128(<4 x float> %{{.*}}, i32 0) 30 // CHECK: extractelement <8 x i16> %{{.*}}, i32 0 31 return _cvtss_sh(a, 0); 32 } 33 34 __m128 test_mm_cvtph_ps(__m128i a) { 35 // CHECK-LABEL: test_mm_cvtph_ps 36 // CHECK: call <4 x float> @llvm.x86.vcvtph2ps.128(<8 x i16> %{{.*}}) 37 return _mm_cvtph_ps(a); 38 } 39 40 __m256 test_mm256_cvtph_ps(__m128i a) { 41 // CHECK-LABEL: test_mm256_cvtph_ps 42 // CHECK: call <8 x float> @llvm.x86.vcvtph2ps.256(<8 x i16> %{{.*}}) 43 return _mm256_cvtph_ps(a); 44 } 45 46 __m128i test_mm_cvtps_ph(__m128 a) { 47 // CHECK-LABEL: test_mm_cvtps_ph 48 // CHECK: call <8 x i16> @llvm.x86.vcvtps2ph.128(<4 x float> %{{.*}}, i32 0) 49 return _mm_cvtps_ph(a, 0); 50 } 51 52 __m128i test_mm256_cvtps_ph(__m256 a) { 53 // CHECK-LABEL: test_mm256_cvtps_ph 54 // CHECK: call <8 x i16> @llvm.x86.vcvtps2ph.256(<8 x float> %{{.*}}, i32 0) 55 return _mm256_cvtps_ph(a, 0); 56 } 57
