This is an archive of the discontinued LLVM Phabricator instance.

Differential D38231

fixing a bug in mask[z]_set1
ClosedPublic

Authored by jina.nahias on Sep 25 2017, 3:45 AM.

Details

Reviewers
craig.topper
zvi
igorb
Commits
rG123c599a0fa4: fixing a bug in mask[z]_set1 intrinsic
rC314102: fixing a bug in mask[z]_set1 intrinsic
rL314102: fixing a bug in mask[z]_set1 intrinsic

Diff Detail

Event Timeline

jina.nahias created this revision.Sep 25 2017, 3:45 AM
jina.nahias updated this revision to Diff 116513.Sep 25 2017, 4:13 AM
igorb accepted this revision.Sep 25 2017, 5:46 AM

LGTM

This revision is now accepted and ready to land.Sep 25 2017, 5:46 AM
Closed by commit rL314102: fixing a bug in mask[z]_set1 intrinsic (authored by jina.nahias). · Explain WhySep 25 2017, 6:39 AM
This revision was automatically updated to reflect the committed changes.

Revision Contents

PathSize
lib/
Headers/
4 lines
test/
CodeGen/
36 lines

Diff 116513

lib/Headers/avx512vlintrin.h

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 5,755 Lines • ▼ Show 20 Lines
} }
#ifdef __x86_64__ #ifdef __x86_64__
static __inline__ __m128i __DEFAULT_FN_ATTRS static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_mask_set1_epi64 (__m128i __O, __mmask8 __M, long long __A) _mm_mask_set1_epi64 (__m128i __O, __mmask8 __M, long long __A)
{ {
return (__m128i) __builtin_ia32_selectq_128(__M, return (__m128i) __builtin_ia32_selectq_128(__M,
(__v2di) _mm_set1_epi8(__A), (__v2di) _mm_set1_epi64x(__A),
(__v2di) __O); (__v2di) __O);
} }
static __inline__ __m128i __DEFAULT_FN_ATTRS static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_maskz_set1_epi64 (__mmask8 __M, long long __A) _mm_maskz_set1_epi64 (__mmask8 __M, long long __A)
{ {
return (__m128i) __builtin_ia32_selectq_128(__M, return (__m128i) __builtin_ia32_selectq_128(__M,
(__v2di) _mm_set1_epi8(__A), (__v2di) _mm_set1_epi64x(__A),
(__v2di) _mm_setzero_si128()); (__v2di) _mm_setzero_si128());
} }
static __inline__ __m256i __DEFAULT_FN_ATTRS static __inline__ __m256i __DEFAULT_FN_ATTRS
_mm256_mask_set1_epi64 (__m256i __O, __mmask8 __M, long long __A) _mm256_mask_set1_epi64 (__m256i __O, __mmask8 __M, long long __A)
{ {
return (__m256i) __builtin_ia32_selectq_256(__M, return (__m256i) __builtin_ia32_selectq_256(__M,
(__v4di) _mm256_set1_epi64x(__A), (__v4di) _mm256_set1_epi64x(__A),
▲ Show 20 Lines • Show All 3,162 Lines • Show Last 20 Lines

test/CodeGen/avx512vl-builtins.c

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 4,557 Lines • ▼ Show 20 Lines__m256i test_mm256_maskz_set1_epi32(__mmask8 __M) {
// CHECK: insertelement <8 x i32> %{{.*}}, i32 %{{.*}}, i32 7 // CHECK: insertelement <8 x i32> %{{.*}}, i32 %{{.*}}, i32 7
// CHECK: select <8 x i1> %{{.*}}, <8 x i32> %{{.*}}, <8 x i32> %{{.*}} // CHECK: select <8 x i1> %{{.*}}, <8 x i32> %{{.*}}, <8 x i32> %{{.*}}
return _mm256_maskz_set1_epi32(__M, 5); return _mm256_maskz_set1_epi32(__M, 5);
} }
#ifdef __x86_64__ #ifdef __x86_64__
__m128i test_mm_mask_set1_epi64(__m128i __O, __mmask8 __M, long long __A) { __m128i test_mm_mask_set1_epi64(__m128i __O, __mmask8 __M, long long __A) {
// CHECK-LABEL: @test_mm_mask_set1_epi64 // CHECK-LABEL: @test_mm_mask_set1_epi64
// CHECK: insertelement <16 x i8> undef, i8 %{{.*}}, i32 0 // CHECK: insertelement <2 x i64> undef, i64 %{{.*}}, i32 0
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 1 // CHECK: insertelement <2 x i64> %{{.*}}, i64 %{{.*}}, i32 1
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 2
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 3
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 4
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 5
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 6
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 7
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 8
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 9
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 10
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 11
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 12
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 13
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 14
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 15
// CHECK: shufflevector <8 x i1> %{{.*}}, <8 x i1> %{{.*}}, <2 x i32> <i32 0, i32 1> // CHECK: shufflevector <8 x i1> %{{.*}}, <8 x i1> %{{.*}}, <2 x i32> <i32 0, i32 1>
// CHECK: select <2 x i1> %{{.*}}, <2 x i64> %{{.*}}, <2 x i64> %{{.*}} // CHECK: select <2 x i1> %{{.*}}, <2 x i64> %{{.*}}, <2 x i64> %{{.*}}
return _mm_mask_set1_epi64(__O, __M, __A); return _mm_mask_set1_epi64(__O, __M, __A);
} }
__m128i test_mm_maskz_set1_epi64(__mmask8 __M, long long __A) { __m128i test_mm_maskz_set1_epi64(__mmask8 __M, long long __A) {
// CHECK-LABEL: @test_mm_maskz_set1_epi64 // CHECK-LABEL: @test_mm_maskz_set1_epi64
// CHECK: insertelement <16 x i8> undef, i8 %{{.*}}, i32 0 // CHECK: insertelement <2 x i64> undef, i64 %{{.*}}, i32 0
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 1 // CHECK: insertelement <2 x i64> %{{.*}}, i64 %{{.*}}, i32 1
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 2
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 3
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 4
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 5
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 6
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 7
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 8
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 9
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 10
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 11
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 12
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 13
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 14
// CHECK: insertelement <16 x i8> %{{.*}}, i8 %{{.*}}, i32 15
// CHECK: shufflevector <8 x i1> %{{.*}}, <8 x i1> %{{.*}}, <2 x i32> <i32 0, i32 1> // CHECK: shufflevector <8 x i1> %{{.*}}, <8 x i1> %{{.*}}, <2 x i32> <i32 0, i32 1>
// CHECK: select <2 x i1> %{{.*}}, <2 x i64> %{{.*}}, <2 x i64> %{{.*}} // CHECK: select <2 x i1> %{{.*}}, <2 x i64> %{{.*}}, <2 x i64> %{{.*}}
return _mm_maskz_set1_epi64(__M, __A); return _mm_maskz_set1_epi64(__M, __A);
} }
__m256i test_mm256_mask_set1_epi64(__m256i __O, __mmask8 __M, long long __A) { __m256i test_mm256_mask_set1_epi64(__m256i __O, __mmask8 __M, long long __A) {
// CHECK-LABEL: @test_mm256_mask_set1_epi64 // CHECK-LABEL: @test_mm256_mask_set1_epi64
// CHECK: insertelement <4 x i64> undef, i64 %{{.*}}, i32 0 // CHECK: insertelement <4 x i64> undef, i64 %{{.*}}, i32 0
▲ Show 20 Lines • Show All 2,817 Lines • Show Last 20 Lines