[clang] a9a0978 - [Headers][X86] Allow MMX/SSE2/AVX2/AVX512BW integer saturated arithmetic intrinsics to be used in constexpr (#153088)
via cfe-commits
cfe-commits at lists.llvm.org
Wed Aug 13 02:27:43 PDT 2025
Author: Yatao Wang
Date: 2025-08-13T10:27:40+01:00
New Revision: a9a0978c1d27d7cbb747520e9333bf450d655569
URL: https://github.com/llvm/llvm-project/commit/a9a0978c1d27d7cbb747520e9333bf450d655569
DIFF: https://github.com/llvm/llvm-project/commit/a9a0978c1d27d7cbb747520e9333bf450d655569.diff
LOG: [Headers][X86] Allow MMX/SSE2/AVX2/AVX512BW integer saturated arithmetic intrinsics to be used in constexpr (#153088)
This PR allows the following MMX/SSE2/AVX2/AVX512BW integer saturated arithmetic intrinsics to be used in constexpr
```
_mm_adds_pi8
_mm_adds_pi16
_mm_subs_pi8
_mm_subs_pi16
_mm_adds_pu8
_mm_adds_pu16
_mm_subs_pu8
_mm_subs_pu16
_mm_adds_epi8
_mm_adds_epi16
_mm_subs_epi8
_mm_subs_epi16
_mm_adds_epu8
_mm_adds_epu16
_mm_subs_epu8
_mm_subs_epu16
_mm256_adds_epi8
_mm256_adds_epi16
_mm256_subs_epi8
_mm256_subs_epi16
_mm256_adds_epu8
_mm256_adds_epu16
_mm256_subs_epu8
_mm256_subs_epu16
_mm512_adds_epi8
_mm512_adds_epi16
_mm512_subs_epi8
_mm512_subs_epi16
_mm512_adds_epu8
_mm512_adds_epu16
_mm512_subs_epu8
_mm512_subs_epu16
```
Fixes #152506
---------
Co-authored-by: Simon Pilgrim <llvm-dev at redking.me.uk>
Added:
Modified:
clang/lib/Headers/avx2intrin.h
clang/lib/Headers/avx512bwintrin.h
clang/lib/Headers/emmintrin.h
clang/lib/Headers/mmintrin.h
clang/test/CodeGen/X86/avx2-builtins.c
clang/test/CodeGen/X86/avx512bw-builtins.c
clang/test/CodeGen/X86/builtin_test_helpers.h
clang/test/CodeGen/X86/mmx-builtins.c
clang/test/CodeGen/X86/sse2-builtins.c
Removed:
################################################################################
diff --git a/clang/lib/Headers/avx2intrin.h b/clang/lib/Headers/avx2intrin.h
index f00a8a516ecfe..35258d47c7188 100644
--- a/clang/lib/Headers/avx2intrin.h
+++ b/clang/lib/Headers/avx2intrin.h
@@ -367,9 +367,8 @@ _mm256_add_epi64(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit integer vector containing one of the source operands.
/// \returns A 256-bit integer vector containing the sums.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_adds_epi8(__m256i __a, __m256i __b)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_adds_epi8(__m256i __a, __m256i __b) {
return (__m256i)__builtin_elementwise_add_sat((__v32qs)__a, (__v32qs)__b);
}
@@ -385,9 +384,8 @@ _mm256_adds_epi8(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit vector of [16 x i16] containing one of the source operands.
/// \returns A 256-bit vector of [16 x i16] containing the sums.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_adds_epi16(__m256i __a, __m256i __b)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_adds_epi16(__m256i __a, __m256i __b) {
return (__m256i)__builtin_elementwise_add_sat((__v16hi)__a, (__v16hi)__b);
}
@@ -404,9 +402,8 @@ _mm256_adds_epi16(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit integer vector containing one of the source operands.
/// \returns A 256-bit integer vector containing the sums.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_adds_epu8(__m256i __a, __m256i __b)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_adds_epu8(__m256i __a, __m256i __b) {
return (__m256i)__builtin_elementwise_add_sat((__v32qu)__a, (__v32qu)__b);
}
@@ -422,9 +419,8 @@ _mm256_adds_epu8(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit vector of [16 x i16] containing one of the source operands.
/// \returns A 256-bit vector of [16 x i16] containing the sums.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_adds_epu16(__m256i __a, __m256i __b)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_adds_epu16(__m256i __a, __m256i __b) {
return (__m256i)__builtin_elementwise_add_sat((__v16hu)__a, (__v16hu)__b);
}
@@ -2607,9 +2603,8 @@ _mm256_sub_epi64(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit integer vector containing the subtrahends.
/// \returns A 256-bit integer vector containing the
diff erences.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_subs_epi8(__m256i __a, __m256i __b)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_subs_epi8(__m256i __a, __m256i __b) {
return (__m256i)__builtin_elementwise_sub_sat((__v32qs)__a, (__v32qs)__b);
}
@@ -2633,9 +2628,8 @@ _mm256_subs_epi8(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit vector of [16 x i16] containing the subtrahends.
/// \returns A 256-bit vector of [16 x i16] containing the
diff erences.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_subs_epi16(__m256i __a, __m256i __b)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_subs_epi16(__m256i __a, __m256i __b) {
return (__m256i)__builtin_elementwise_sub_sat((__v16hi)__a, (__v16hi)__b);
}
@@ -2660,9 +2654,8 @@ _mm256_subs_epi16(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit integer vector containing the subtrahends.
/// \returns A 256-bit integer vector containing the
diff erences.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_subs_epu8(__m256i __a, __m256i __b)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_subs_epu8(__m256i __a, __m256i __b) {
return (__m256i)__builtin_elementwise_sub_sat((__v32qu)__a, (__v32qu)__b);
}
@@ -2686,9 +2679,8 @@ _mm256_subs_epu8(__m256i __a, __m256i __b)
/// \param __b
/// A 256-bit vector of [16 x i16] containing the subtrahends.
/// \returns A 256-bit vector of [16 x i16] containing the
diff erences.
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_subs_epu16(__m256i __a, __m256i __b)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_subs_epu16(__m256i __a, __m256i __b) {
return (__m256i)__builtin_elementwise_sub_sat((__v16hu)__a, (__v16hu)__b);
}
diff --git a/clang/lib/Headers/avx512bwintrin.h b/clang/lib/Headers/avx512bwintrin.h
index d7377e5d5196c..886783230cf32 100644
--- a/clang/lib/Headers/avx512bwintrin.h
+++ b/clang/lib/Headers/avx512bwintrin.h
@@ -613,9 +613,8 @@ _mm512_maskz_packus_epi16(__mmask64 __M, __m512i __A, __m512i __B)
(__v64qi)_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_adds_epi8 (__m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_adds_epi8(__m512i __A, __m512i __B) {
return (__m512i)__builtin_elementwise_add_sat((__v64qs)__A, (__v64qs)__B);
}
@@ -635,9 +634,8 @@ _mm512_maskz_adds_epi8 (__mmask64 __U, __m512i __A, __m512i __B)
(__v64qi)_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_adds_epi16 (__m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_adds_epi16(__m512i __A, __m512i __B) {
return (__m512i)__builtin_elementwise_add_sat((__v32hi)__A, (__v32hi)__B);
}
@@ -657,9 +655,8 @@ _mm512_maskz_adds_epi16 (__mmask32 __U, __m512i __A, __m512i __B)
(__v32hi)_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_adds_epu8 (__m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_adds_epu8(__m512i __A, __m512i __B) {
return (__m512i)__builtin_elementwise_add_sat((__v64qu) __A, (__v64qu) __B);
}
@@ -679,9 +676,8 @@ _mm512_maskz_adds_epu8 (__mmask64 __U, __m512i __A, __m512i __B)
(__v64qi)_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_adds_epu16 (__m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_adds_epu16(__m512i __A, __m512i __B) {
return (__m512i)__builtin_elementwise_add_sat((__v32hu) __A, (__v32hu) __B);
}
@@ -946,9 +942,8 @@ _mm512_maskz_shuffle_epi8(__mmask64 __U, __m512i __A, __m512i __B)
(__v64qi)_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_subs_epi8 (__m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_subs_epi8(__m512i __A, __m512i __B) {
return (__m512i)__builtin_elementwise_sub_sat((__v64qs)__A, (__v64qs)__B);
}
@@ -968,9 +963,8 @@ _mm512_maskz_subs_epi8 (__mmask64 __U, __m512i __A, __m512i __B)
(__v64qi)_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_subs_epi16 (__m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_subs_epi16(__m512i __A, __m512i __B) {
return (__m512i)__builtin_elementwise_sub_sat((__v32hi)__A, (__v32hi)__B);
}
@@ -990,9 +984,8 @@ _mm512_maskz_subs_epi16 (__mmask32 __U, __m512i __A, __m512i __B)
(__v32hi)_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_subs_epu8 (__m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_subs_epu8(__m512i __A, __m512i __B) {
return (__m512i)__builtin_elementwise_sub_sat((__v64qu) __A, (__v64qu) __B);
}
@@ -1012,9 +1005,8 @@ _mm512_maskz_subs_epu8 (__mmask64 __U, __m512i __A, __m512i __B)
(__v64qi)_mm512_setzero_si512());
}
-static __inline__ __m512i __DEFAULT_FN_ATTRS512
-_mm512_subs_epu16 (__m512i __A, __m512i __B)
-{
+static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR
+_mm512_subs_epu16(__m512i __A, __m512i __B) {
return (__m512i)__builtin_elementwise_sub_sat((__v32hu) __A, (__v32hu) __B);
}
diff --git a/clang/lib/Headers/emmintrin.h b/clang/lib/Headers/emmintrin.h
index e632191113c4a..1973ccb0e00bf 100644
--- a/clang/lib/Headers/emmintrin.h
+++ b/clang/lib/Headers/emmintrin.h
@@ -2170,8 +2170,8 @@ _mm_add_epi64(__m128i __a, __m128i __b) {
/// A 128-bit signed [16 x i8] vector.
/// \returns A 128-bit signed [16 x i8] vector containing the saturated sums of
/// both parameters.
-static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_adds_epi8(__m128i __a,
- __m128i __b) {
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_adds_epi8(__m128i __a, __m128i __b) {
return (__m128i)__builtin_elementwise_add_sat((__v16qs)__a, (__v16qs)__b);
}
@@ -2192,8 +2192,8 @@ static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_adds_epi8(__m128i __a,
/// A 128-bit signed [8 x i16] vector.
/// \returns A 128-bit signed [8 x i16] vector containing the saturated sums of
/// both parameters.
-static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_adds_epi16(__m128i __a,
- __m128i __b) {
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_adds_epi16(__m128i __a, __m128i __b) {
return (__m128i)__builtin_elementwise_add_sat((__v8hi)__a, (__v8hi)__b);
}
@@ -2214,8 +2214,8 @@ static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_adds_epi16(__m128i __a,
/// A 128-bit unsigned [16 x i8] vector.
/// \returns A 128-bit unsigned [16 x i8] vector containing the saturated sums
/// of both parameters.
-static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_adds_epu8(__m128i __a,
- __m128i __b) {
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_adds_epu8(__m128i __a, __m128i __b) {
return (__m128i)__builtin_elementwise_add_sat((__v16qu)__a, (__v16qu)__b);
}
@@ -2236,8 +2236,8 @@ static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_adds_epu8(__m128i __a,
/// A 128-bit unsigned [8 x i16] vector.
/// \returns A 128-bit unsigned [8 x i16] vector containing the saturated sums
/// of both parameters.
-static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_adds_epu16(__m128i __a,
- __m128i __b) {
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_adds_epu16(__m128i __a, __m128i __b) {
return (__m128i)__builtin_elementwise_add_sat((__v8hu)__a, (__v8hu)__b);
}
@@ -2597,8 +2597,8 @@ _mm_sub_epi64(__m128i __a, __m128i __b) {
/// A 128-bit integer vector containing the subtrahends.
/// \returns A 128-bit integer vector containing the
diff erences of the values
/// in the operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_subs_epi8(__m128i __a,
- __m128i __b) {
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_subs_epi8(__m128i __a, __m128i __b) {
return (__m128i)__builtin_elementwise_sub_sat((__v16qs)__a, (__v16qs)__b);
}
@@ -2619,8 +2619,8 @@ static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_subs_epi8(__m128i __a,
/// A 128-bit integer vector containing the subtrahends.
/// \returns A 128-bit integer vector containing the
diff erences of the values
/// in the operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_subs_epi16(__m128i __a,
- __m128i __b) {
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_subs_epi16(__m128i __a, __m128i __b) {
return (__m128i)__builtin_elementwise_sub_sat((__v8hi)__a, (__v8hi)__b);
}
@@ -2640,8 +2640,8 @@ static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_subs_epi16(__m128i __a,
/// A 128-bit integer vector containing the subtrahends.
/// \returns A 128-bit integer vector containing the unsigned integer
///
diff erences of the values in the operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_subs_epu8(__m128i __a,
- __m128i __b) {
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_subs_epu8(__m128i __a, __m128i __b) {
return (__m128i)__builtin_elementwise_sub_sat((__v16qu)__a, (__v16qu)__b);
}
@@ -2661,8 +2661,8 @@ static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_subs_epu8(__m128i __a,
/// A 128-bit integer vector containing the subtrahends.
/// \returns A 128-bit integer vector containing the unsigned integer
///
diff erences of the values in the operands.
-static __inline__ __m128i __DEFAULT_FN_ATTRS _mm_subs_epu16(__m128i __a,
- __m128i __b) {
+static __inline__ __m128i __DEFAULT_FN_ATTRS_CONSTEXPR
+_mm_subs_epu16(__m128i __a, __m128i __b) {
return (__m128i)__builtin_elementwise_sub_sat((__v8hu)__a, (__v8hu)__b);
}
diff --git a/clang/lib/Headers/mmintrin.h b/clang/lib/Headers/mmintrin.h
index 3961b790cea74..f62450bb1a67f 100644
--- a/clang/lib/Headers/mmintrin.h
+++ b/clang/lib/Headers/mmintrin.h
@@ -448,10 +448,9 @@ _mm_add_pi32(__m64 __m1, __m64 __m2)
/// A 64-bit integer vector of [8 x i8].
/// \returns A 64-bit integer vector of [8 x i8] containing the saturated sums
/// of both parameters.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_adds_pi8(__m64 __m1, __m64 __m2)
-{
- return (__m64)__builtin_elementwise_add_sat((__v8qs)__m1, (__v8qs)__m2);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_adds_pi8(__m64 __m1, __m64 __m2) {
+ return (__m64)__builtin_elementwise_add_sat((__v8qs)__m1, (__v8qs)__m2);
}
/// Adds, with saturation, each 16-bit signed integer element of the first
@@ -472,10 +471,9 @@ _mm_adds_pi8(__m64 __m1, __m64 __m2)
/// A 64-bit integer vector of [4 x i16].
/// \returns A 64-bit integer vector of [4 x i16] containing the saturated sums
/// of both parameters.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_adds_pi16(__m64 __m1, __m64 __m2)
-{
- return (__m64)__builtin_elementwise_add_sat((__v4hi)__m1, (__v4hi)__m2);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_adds_pi16(__m64 __m1, __m64 __m2) {
+ return (__m64)__builtin_elementwise_add_sat((__v4hi)__m1, (__v4hi)__m2);
}
/// Adds, with saturation, each 8-bit unsigned integer element of the first
@@ -495,10 +493,9 @@ _mm_adds_pi16(__m64 __m1, __m64 __m2)
/// A 64-bit integer vector of [8 x i8].
/// \returns A 64-bit integer vector of [8 x i8] containing the saturated
/// unsigned sums of both parameters.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_adds_pu8(__m64 __m1, __m64 __m2)
-{
- return (__m64)__builtin_elementwise_add_sat((__v8qu)__m1, (__v8qu)__m2);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_adds_pu8(__m64 __m1, __m64 __m2) {
+ return (__m64)__builtin_elementwise_add_sat((__v8qu)__m1, (__v8qu)__m2);
}
/// Adds, with saturation, each 16-bit unsigned integer element of the first
@@ -518,10 +515,9 @@ _mm_adds_pu8(__m64 __m1, __m64 __m2)
/// A 64-bit integer vector of [4 x i16].
/// \returns A 64-bit integer vector of [4 x i16] containing the saturated
/// unsigned sums of both parameters.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_adds_pu16(__m64 __m1, __m64 __m2)
-{
- return (__m64)__builtin_elementwise_add_sat((__v4hu)__m1, (__v4hu)__m2);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_adds_pu16(__m64 __m1, __m64 __m2) {
+ return (__m64)__builtin_elementwise_add_sat((__v4hu)__m1, (__v4hu)__m2);
}
/// Subtracts each 8-bit integer element of the second 64-bit integer
@@ -605,10 +601,9 @@ _mm_sub_pi32(__m64 __m1, __m64 __m2)
/// A 64-bit integer vector of [8 x i8] containing the subtrahends.
/// \returns A 64-bit integer vector of [8 x i8] containing the saturated
///
diff erences of both parameters.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_subs_pi8(__m64 __m1, __m64 __m2)
-{
- return (__m64)__builtin_elementwise_sub_sat((__v8qs)__m1, (__v8qs)__m2);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_subs_pi8(__m64 __m1, __m64 __m2) {
+ return (__m64)__builtin_elementwise_sub_sat((__v8qs)__m1, (__v8qs)__m2);
}
/// Subtracts, with saturation, each 16-bit signed integer element of the
@@ -629,10 +624,9 @@ _mm_subs_pi8(__m64 __m1, __m64 __m2)
/// A 64-bit integer vector of [4 x i16] containing the subtrahends.
/// \returns A 64-bit integer vector of [4 x i16] containing the saturated
///
diff erences of both parameters.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_subs_pi16(__m64 __m1, __m64 __m2)
-{
- return (__m64)__builtin_elementwise_sub_sat((__v4hi)__m1, (__v4hi)__m2);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_subs_pi16(__m64 __m1, __m64 __m2) {
+ return (__m64)__builtin_elementwise_sub_sat((__v4hi)__m1, (__v4hi)__m2);
}
/// Subtracts each 8-bit unsigned integer element of the second 64-bit
@@ -653,10 +647,9 @@ _mm_subs_pi16(__m64 __m1, __m64 __m2)
/// A 64-bit integer vector of [8 x i8] containing the subtrahends.
/// \returns A 64-bit integer vector of [8 x i8] containing the saturated
///
diff erences of both parameters.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_subs_pu8(__m64 __m1, __m64 __m2)
-{
- return (__m64)__builtin_elementwise_sub_sat((__v8qu)__m1, (__v8qu)__m2);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_subs_pu8(__m64 __m1, __m64 __m2) {
+ return (__m64)__builtin_elementwise_sub_sat((__v8qu)__m1, (__v8qu)__m2);
}
/// Subtracts each 16-bit unsigned integer element of the second 64-bit
@@ -677,10 +670,9 @@ _mm_subs_pu8(__m64 __m1, __m64 __m2)
/// A 64-bit integer vector of [4 x i16] containing the subtrahends.
/// \returns A 64-bit integer vector of [4 x i16] containing the saturated
///
diff erences of both parameters.
-static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2
-_mm_subs_pu16(__m64 __m1, __m64 __m2)
-{
- return (__m64)__builtin_elementwise_sub_sat((__v4hu)__m1, (__v4hu)__m2);
+static __inline__ __m64 __DEFAULT_FN_ATTRS_SSE2_CONSTEXPR
+_mm_subs_pu16(__m64 __m1, __m64 __m2) {
+ return (__m64)__builtin_elementwise_sub_sat((__v4hu)__m1, (__v4hu)__m2);
}
/// Multiplies each 16-bit signed integer element of the first 64-bit
diff --git a/clang/test/CodeGen/X86/avx2-builtins.c b/clang/test/CodeGen/X86/avx2-builtins.c
index 8790485f00a8c..76d26d57cfecf 100644
--- a/clang/test/CodeGen/X86/avx2-builtins.c
+++ b/clang/test/CodeGen/X86/avx2-builtins.c
@@ -60,12 +60,14 @@ __m256i test_mm256_adds_epi8(__m256i a, __m256i b) {
// CHECK: call <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8> %{{.*}}, <32 x i8> %{{.*}})
return _mm256_adds_epi8(a, b);
}
+TEST_CONSTEXPR(match_v32qi(_mm256_adds_epi8((__m256i)(__v32qs){0, +1, +2, +3, +4, +5, +6, +7, +8, +9, +10, +11, +12, +13, +14, +15, +100, +50, -100, +20, +80, -50, +120, -20, -100, -50, +100, -20, -80, +50, -120, +20}, (__m256i)(__v32qs){0, +1, +2, +3, +4, +5, +6, +7, +8, +9, +10, +11, +12, +13, +14, +15, +50, +80, -50, +110, +60, -30, +20, -10, +50, +80, -50, +110, +60, -30, +20, -10}), 0, +2, +4, +6, +8, +10, +12, +14, +16, +18, +20, +22, +24, +26, +28, +30, +127, +127, -128, +127, +127, -80, +127, -30, -50, +30, +50, +90, -20, +20, -100, +10));
__m256i test_mm256_adds_epi16(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_adds_epi16
// CHECK: call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> %{{.*}}, <16 x i16> %{{.*}})
return _mm256_adds_epi16(a, b);
}
+TEST_CONSTEXPR(match_v16hi(_mm256_adds_epi16((__m256i)(__v16hi){0, -1, -2, -3, -4, -5, -6, -7, -8, -9, -10, -11, +32000, -32000, +32000, -32000}, (__m256i)(__v16hi){0, -1, -2, -3, -4, -5, -6, -7, -8, -9, -10, -11, +800, -800, -800, +800}), 0, -2, -4, -6, -8, -10, -12, -14, -16, -18, -20, -22, +32767, -32768, +31200, -31200));
__m256i test_mm256_adds_epu8(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_adds_epu8
@@ -73,6 +75,7 @@ __m256i test_mm256_adds_epu8(__m256i a, __m256i b) {
// CHECK: call <32 x i8> @llvm.uadd.sat.v32i8(<32 x i8> %{{.*}}, <32 x i8> %{{.*}})
return _mm256_adds_epu8(a, b);
}
+TEST_CONSTEXPR(match_v32qu(_mm256_adds_epu8((__m256i)(__v32qu){0, 0, 0, 0, +64, +64, +64, +64, +64, +64, +127, +127, +127, +127, +127, +127, +128, +128, +128, +128, +128, +128, +192, +192, +192, +192, +192, +192, +255, +255, +255, +255}, (__m256i)(__v32qu){0, +127, +128, +255, 0, +64, +127, +128, +192, +255, 0, +64, +127, +128, +192, +255, 0, +64, +127, +128, +192, +255, 0, +64, +127, +128, +192, +255, 0, +127, +128, +255}), 0, +127, +128, +255, +64, +128, +191, +192, +255, +255, +127, +191, +254, +255, +255, +255, +128, +192, +255, +255, +255, +255, +192, +255, +255, +255, +255, +255, +255, +255, +255, +255));
__m256i test_mm256_adds_epu16(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_adds_epu16
@@ -80,6 +83,7 @@ __m256i test_mm256_adds_epu16(__m256i a, __m256i b) {
// CHECK: call <16 x i16> @llvm.uadd.sat.v16i16(<16 x i16> %{{.*}}, <16 x i16> %{{.*}})
return _mm256_adds_epu16(a, b);
}
+TEST_CONSTEXPR(match_v16hu(_mm256_adds_epu16((__m256i)(__v16hu){0, 0, 0, 0, +32767, +32767, +32767, +32767, +32768, +32768, +32768, +32768, +65535, +65535, +65535, +65535}, (__m256i)(__v16hu){0, +32767, +32768, +65535, 0, +32767, +32768, +65535, 0, +32767, +32768, +65535, 0, +32767, +32768, +65535}), 0, +32767, +32768, +65535, +32767, +65534, +65535, +65535, +32768, +65535, +65535, +65535, +65535, +65535, +65535, +65535));
__m256i test_mm256_alignr_epi8(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_alignr_epi8
@@ -1293,12 +1297,14 @@ __m256i test_mm256_subs_epi8(__m256i a, __m256i b) {
// CHECK: call <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8> %{{.*}}, <32 x i8> %{{.*}})
return _mm256_subs_epi8(a, b);
}
+TEST_CONSTEXPR(match_v32qi(_mm256_subs_epi8((__m256i)(__v32qs){0, +1, +2, +3, +4, +5, +6, +7, +8, +9, +10, +11, +12, +13, +14, +15, +100, +50, -100, +20, +80, -50, +120, -20, -100, -50, +100, -20, -80, +50, -120, +20}, (__m256i)(__v32qs){0, -1, -2, -3, -4, -5, -6, -7, -8, -9, -10, -11, -12, -13, -14, -15, -50, -80, +50, -110, -60, +30, -20, +10, -50, -80, +50, -110, -60, +30, -20, +10}), 0, +2, +4, +6, +8, +10, +12, +14, +16, +18, +20, +22, +24, +26, +28, +30, +127, +127, -128, +127, +127, -80, +127, -30, -50, +30, +50, +90, -20, +20, -100, +10));
__m256i test_mm256_subs_epi16(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_subs_epi16
// CHECK: call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> %{{.*}}, <16 x i16> %{{.*}})
return _mm256_subs_epi16(a, b);
}
+TEST_CONSTEXPR(match_v16hi(_mm256_subs_epi16((__m256i)(__v16hi){0, -1, -2, -3, -4, -5, -6, -7, -8, -9, -10, -11, +32000, -32000, +32000, -32000}, (__m256i)(__v16hi){0, +1, +2, +3, +4, +5, +6, +7, +8, +9, +10, +11, -800, +800, +800, -800}),0, -2, -4, -6, -8, -10, -12, -14, -16, -18, -20, -22, +32767, -32768, +31200, -31200));
__m256i test_mm256_subs_epu8(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_subs_epu8
@@ -1306,6 +1312,7 @@ __m256i test_mm256_subs_epu8(__m256i a, __m256i b) {
// CHECK: call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> %{{.*}}, <32 x i8> %{{.*}})
return _mm256_subs_epu8(a, b);
}
+TEST_CONSTEXPR(match_v32qu(_mm256_subs_epu8((__m256i)(__v32qu){0, 0, 0, 0, +64, +64, +64, +64, +64, +64, +127, +127, +127, +127, +127, +127, +128, +128, +128, +128, +128, +128, +192, +192, +192, +192, +192, +192, +255, +255, +255, +255}, (__m256i)(__v32qu){0, +127, +128, +255, 0, +64, +127, +128, +192, +255, 0, +64, +127, +128, +192, +255, 0, +64, +127, +128, +192, +255, 0, +64, +127, +128, +192, +255, 0, +127, +128, +255}), 0, 0, 0, 0, +64, 0, 0, 0, 0, 0, +127, +63, 0, 0, 0, 0, +128, +64, +1, 0, 0, 0, +192, +128, +65, +64, 0, 0, +255, +128, +127, 0));
__m256i test_mm256_subs_epu16(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_subs_epu16
@@ -1313,6 +1320,7 @@ __m256i test_mm256_subs_epu16(__m256i a, __m256i b) {
// CHECK: call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> %{{.*}}, <16 x i16> %{{.*}})
return _mm256_subs_epu16(a, b);
}
+TEST_CONSTEXPR(match_v16hu(_mm256_subs_epu16((__m256i)(__v16hu){0, 0, 0, 0, +32767, +32767, +32767, +32767, +32768, +32768, +32768, +32768, +65535, +65535, +65535, +65535}, (__m256i)(__v16hu){0, +32767, +32768, +65535, 0, +32767, +32768, +65535, 0, +32767, +32768, +65535, 0, +32767, +32768, +65535}), 0, 0, 0, 0, +32767, 0, 0, 0, +32768, +1, 0, 0, +65535, +32768, +32767, 0));
__m256i test_mm256_unpackhi_epi8(__m256i a, __m256i b) {
// CHECK-LABEL: test_mm256_unpackhi_epi8
diff --git a/clang/test/CodeGen/X86/avx512bw-builtins.c b/clang/test/CodeGen/X86/avx512bw-builtins.c
index 2e3e13d2769c4..67f7df03469b6 100644
--- a/clang/test/CodeGen/X86/avx512bw-builtins.c
+++ b/clang/test/CodeGen/X86/avx512bw-builtins.c
@@ -957,6 +957,8 @@ __m512i test_mm512_adds_epi8(__m512i __A, __m512i __B) {
// CHECK: @llvm.sadd.sat.v64i8
return _mm512_adds_epi8(__A,__B);
}
+TEST_CONSTEXPR(match_v64qi(_mm512_adds_epi8((__m512i)(__v64qi){0, +1, -2, +3, -4, +5, -6, +7, -8, +9, -10, +11, -12, +13, -14, +15, -16, +17, -18, +19, -20, +21, -22, +23, -24, +25, -26, +27, -28, +29, -30, +31, -32, +33, -34, +35, -36, +37, -38, +39, -40, +41, -42, +43, -44, +45, -46, +47, +100, +50, -100, +20, +80, -50, +120, -20, -100, -50, +100, -20, -80, +50, -120, +20}, (__m512i)(__v64qi){0, +1, -2, +3, -4, +5, -6, +7, -8, +9, -10, +11, -12, +13, -14, +15, -16, +17, -18, +19, -20, +21, -22, +23, -24, +25, -26, +27, -28, +29, -30, +31, -32, +33, -34, +35, -36, +37, -38, +39, -40, +41, -42, +43, -44, +45, -46, +47, +50, +80, -50, +110, +60, -30, +20, -10, +50, +80, -50, +110, +60, -30, +20, -10}), 0, +2, -4, +6, -8, +10, -12, +14, -16, +18, -20, +22, -24, +26, -28, +30, -32, +34, -36, +38, -40, +42, -44, +46, -48, +50, -52, +54, -56, +58, -60, +62, -64, +66, -68, +70, -72, +74, -76, +78, -80, +82, -84, +86, -88, +90, -92, +94, +127, +127, -128, +127, +127, -80, +127, -30, -50, +30, +50, +90, -20, +20, -100, +10));
+
__m512i test_mm512_mask_adds_epi8(__m512i __W, __mmask64 __U, __m512i __A, __m512i __B) {
// CHECK-LABEL: @test_mm512_mask_adds_epi8
// CHECK: @llvm.sadd.sat.v64i8
@@ -974,6 +976,8 @@ __m512i test_mm512_adds_epi16(__m512i __A, __m512i __B) {
// CHECK: @llvm.sadd.sat.v32i16
return _mm512_adds_epi16(__A,__B);
}
+TEST_CONSTEXPR(match_v32hi(_mm512_adds_epi16((__m512i)(__v32hi){0, +1, -2, +3, -4, +5, -6, +7, -8, +9, -10, +11, -12, +13, -14, +15, -16, +17, -18, +19, -20, +21, -22, +23, -24, +25, -26, +27, +32000, -32000, +32000, -32000}, (__m512i)(__v32hi){0, +1, -2, +3, -4, +5, -6, +7, -8, +9, -10, +11, -12, +13, -14, +15, -16, +17, -18, +19, -20, +21, -22, +23, -24, +25, -26, +27, +800, -800, -800, +800}), 0, +2, -4, +6, -8, +10, -12, +14, -16, +18, -20, +22, -24, +26, -28, +30, -32, +34, -36, +38, -40, +42, -44, +46, -48, +50, -52, +54, +32767, -32768, +31200, -31200));
+
__m512i test_mm512_mask_adds_epi16(__m512i __W, __mmask32 __U, __m512i __A, __m512i __B) {
// CHECK-LABEL: @test_mm512_mask_adds_epi16
// CHECK: @llvm.sadd.sat.v32i16
@@ -992,6 +996,8 @@ __m512i test_mm512_adds_epu8(__m512i __A, __m512i __B) {
// CHECK: call <64 x i8> @llvm.uadd.sat.v64i8(<64 x i8> %{{.*}}, <64 x i8> %{{.*}})
return _mm512_adds_epu8(__A,__B);
}
+TEST_CONSTEXPR(match_v64qu(_mm512_adds_epu8((__m512i)(__v64qu){0, 0, 0, 0, 0, 0, 0, 0, +63, +63, +63, +63, +63, +63, +63, +63, +64, +64, +64, +64, +64, +64, +64, +64, +127, +127, +127, +127, +127, +127, +127, +127, +128, +128, +128, +128, +128, +128, +128, +128, +191, +191, +191, +191, +191, +191, +191, +191, +192, +192, +192, +192, +192, +192, +192, +192, +255, +255, +255, +255, +255, +255, +255, +255}, (__m512i)(__v64qu){0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255}), 0, +63, +64, +127, +128, +191, +192, +255, +63, +126, +127, +190, +191, +254, +255, +255, +64, +127, +128, +192, +193, +255, +255, +255, +127, +190, +191, +254, +255, +255, +255, +255, +128, +191, +192, +255, +255, +255, +255, +255, +191, +254, +255, +255, +255, +255, +255, +255, +192, +255, +255, +255, +255, +255, +255, +255, +255, +255, +255, +255, +255, +255, +255, +255));
+
__m512i test_mm512_mask_adds_epu8(__m512i __W, __mmask64 __U, __m512i __A, __m512i __B) {
// CHECK-LABEL: @test_mm512_mask_adds_epu8
// CHECK-NOT: @llvm.x86.avx512.mask.paddus.b.512
@@ -999,6 +1005,8 @@ __m512i test_mm512_mask_adds_epu8(__m512i __W, __mmask64 __U, __m512i __A, __m51
// CHECK: select <64 x i1> %{{.*}}, <64 x i8> %{{.*}}, <64 x i8> %{{.*}}
return _mm512_mask_adds_epu8(__W,__U,__A,__B);
}
+TEST_CONSTEXPR(match_v32hu(_mm512_adds_epu16((__m512i)(__v32hu){0, 0, 0, 0, +16384, +16384, +16384, +16384, +16384, +16384, +32767, +32767, +32767, +32767, +32767, +32767, +32768, +32768, +32768, +32768, +32768, +32768, +49152, +49152, +49152, +49152, +49152, +49152, +65535, +65535, +65535, +65535}, (__m512i)(__v32hu){0, +32767, +32768, +65535, 0, +16384, +32767, +32768, +49152, +65535, 0, +16384, +32767, +32768, +49152, +65535, 0, +16384, +32767, +32768, +49152, +65535, 0, +16384, +32767, +32768, +49152, +65535, 0, +32767, +32768, +65535}), 0, +32767, +32768, +65535, +16384, +32768, +49151, +49152, +65535, +65535, +32767, +49151, +65534, +65535, +65535, +65535, +32768, +49152, +65535, +65535, +65535, +65535, +49152, +65535, +65535, +65535, +65535, +65535, +65535, +65535, +65535, +65535));
+
__m512i test_mm512_maskz_adds_epu8(__mmask64 __U, __m512i __A, __m512i __B) {
// CHECK-LABEL: @test_mm512_maskz_adds_epu8
// CHECK-NOT: @llvm.x86.avx512.mask.paddus.b.512
@@ -1218,6 +1226,8 @@ __m512i test_mm512_subs_epi8(__m512i __A, __m512i __B) {
// CHECK: @llvm.ssub.sat.v64i8
return _mm512_subs_epi8(__A,__B);
}
+TEST_CONSTEXPR(match_v64qi(_mm512_subs_epi8((__m512i)(__v64qi){0, +1, -2, +3, -4, +5, -6, +7, -8, +9, -10, +11, -12, +13, -14, +15, -16, +17, -18, +19, -20, +21, -22, +23, -24, +25, -26, +27, -28, +29, -30, +31, -32, +33, -34, +35, -36, +37, -38, +39, -40, +41, -42, +43, -44, +45, -46, +47, +100, +50, -100, +20, +80, -50, +120, -20, -100, -50, +100, -20, -80, +50, -120, +20}, (__m512i)(__v64qi){0, -1, +2, -3, +4, -5, +6, -7, +8, -9, +10, -11, +12, -13, +14, -15, +16, -17, +18, -19, +20, -21, +22, -23, +24, -25, +26, -27, +28, -29, +30, -31, +32, -33, +34, -35, +36, -37, +38, -39, +40, -41, +42, -43, +44, -45, +46, -47, -50, -80, +50, -110, -60, +30, -20, +10, -50, -80, +50, -110, -60, +30, -20, +10}), 0, +2, -4, +6, -8, +10, -12, +14, -16, +18, -20, +22, -24, +26, -28, +30, -32, +34, -36, +38, -40, +42, -44, +46, -48, +50, -52, +54, -56, +58, -60, +62, -64, +66, -68, +70, -72, +74, -76, +78, -80, +82, -84, +86, -88, +90, -92, +94, +127, +127, -128, +127, +127, -80, +127, -30, -50, +30, +50, +90, -20, +20, -100, +10));
+
__m512i test_mm512_mask_subs_epi8(__m512i __W, __mmask64 __U, __m512i __A, __m512i __B) {
// CHECK-LABEL: @test_mm512_mask_subs_epi8
// CHECK: @llvm.ssub.sat.v64i8
@@ -1235,6 +1245,7 @@ __m512i test_mm512_subs_epi16(__m512i __A, __m512i __B) {
// CHECK: @llvm.ssub.sat.v32i16
return _mm512_subs_epi16(__A,__B);
}
+TEST_CONSTEXPR(match_v32hi(_mm512_subs_epi16((__m512i)(__v32hi){0, +1, -2, +3, -4, +5, -6, +7, -8, +9, -10, +11, -12, +13, -14, +15, -16, +17, -18, +19, -20, +21, -22, +23, -24, +25, -26, +27, +32000, -32000, +32000, -32000}, (__m512i)(__v32hi){0, -1, +2, -3, +4, -5, +6, -7, +8, -9, +10, -11, +12, -13, +14, -15, +16, -17, +18, -19, +20, -21, +22, -23, +24, -25, +26, -27, -800, +800, +800, -800}), 0, +2, -4, +6, -8, +10, -12, +14, -16, +18, -20, +22, -24, +26, -28, +30, -32, +34, -36, +38, -40, +42, -44, +46, -48, +50, -52, +54, +32767, -32768, +31200, -31200));
__m512i test_mm512_mask_subs_epi16(__m512i __W, __mmask32 __U, __m512i __A, __m512i __B) {
// CHECK-LABEL: @test_mm512_mask_subs_epi16
// CHECK: @llvm.ssub.sat.v32i16
@@ -1260,6 +1271,8 @@ __m512i test_mm512_mask_subs_epu8(__m512i __W, __mmask64 __U, __m512i __A, __m51
// CHECK: select <64 x i1> %{{.*}}, <64 x i8> %{{.*}}, <64 x i8> %{{.*}}
return _mm512_mask_subs_epu8(__W,__U,__A,__B);
}
+TEST_CONSTEXPR(match_v64qu(_mm512_subs_epu8((__m512i)(__v64qu){0, 0, 0, 0, 0, 0, 0, 0, +63, +63, +63, +63, +63, +63, +63, +63, +64, +64, +64, +64, +64, +64, +64, +64, +127, +127, +127, +127, +127, +127, +127, +127, +128, +128, +128, +128, +128, +128, +128, +128, +191, +191, +191, +191, +191, +191, +191, +191, +192, +192, +192, +192, +192, +192, +192, +192, +255, +255, +255, +255, +255, +255, +255, +255}, (__m512i)(__v64qu){0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255, 0, +63, +64, +127, +128, +191, +192, +255}), 0, 0, 0, 0, 0, 0, 0, 0, +63, 0, 0, 0, 0, 0, 0, 0, +64, +1, 0, 0, 0, 0, 0, 0, +127, +64, +63, 0, 0, 0, 0, 0, +128, +65, +64, +1, 0, 0, 0, 0, +191, +128, +127, +64, +63, 0, 0, 0, +192, +129, +128, +65, +64, +1, 0, 0, +255, +192, +191, +128, +127, +64, +63, +0));
+
__m512i test_mm512_maskz_subs_epu8(__mmask64 __U, __m512i __A, __m512i __B) {
// CHECK-LABEL: @test_mm512_maskz_subs_epu8
// CHECK-NOT: @llvm.x86.avx512.mask.psubus.b.512
@@ -1279,6 +1292,7 @@ __m512i test_mm512_mask_subs_epu16(__m512i __W, __mmask32 __U, __m512i __A, __m5
// CHECK: call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> %{{.*}}, <32 x i16> %{{.*}})
// CHECK: select <32 x i1> %{{.*}}, <32 x i16> %{{.*}}, <32 x i16> %{{.*}}
return _mm512_mask_subs_epu16(__W,__U,__A,__B);
+TEST_CONSTEXPR(match_v32hu(_mm512_subs_epu16((__m512i)(__v32hu){0, 0, 0, 0, +16384, +16384, +16384, +16384, +16384, +16384, +32767, +32767, +32767, +32767, +32767, +32767, +32768, +32768, +32768, +32768, +32768, +32768, +49152, +49152, +49152, +49152, +49152, +49152, +65535, +65535, +65535, +65535}, (__m512i)(__v32hu){0, +32767, +32768, +65535, 0, +16384, +32767, +32768, +49152, +65535, 0, +16384, +32767, +32768, +49152, +65535, 0, +16384, +32767, +32768, +49152, +65535, 0, +16384, +32767, +32768, +49152, +65535, 0, +32767, +32768, +65535}), 0, 0, 0, 0, +16384, 0, 0, 0, 0, 0, +32767, +16383, 0, 0, 0, 0, +32768, +16384, +1, 0, 0, 0, +49152, +32768, +16385, +16384, 0, 0, +65535, +32768, +32767, 0));
}
__m512i test_mm512_maskz_subs_epu16(__mmask32 __U, __m512i __A, __m512i __B) {
// CHECK-LABEL: @test_mm512_maskz_subs_epu16
diff --git a/clang/test/CodeGen/X86/builtin_test_helpers.h b/clang/test/CodeGen/X86/builtin_test_helpers.h
index 21028443426ec..b83ca4d711e9c 100644
--- a/clang/test/CodeGen/X86/builtin_test_helpers.h
+++ b/clang/test/CodeGen/X86/builtin_test_helpers.h
@@ -23,11 +23,21 @@ constexpr bool match_v4hi(__m64 _v, short a, short b, short c, short d) {
return v[0] == a && v[1] == b && v[2] == c && v[3] == d;
}
+constexpr bool match_v4hu(__m64 _v, unsigned short a, unsigned short b, unsigned short c, unsigned short d) {
+ __v4hu v = (__v4hu)_v;
+ return v[0] == a && v[1] == b && v[2] == c && v[3] == d;
+}
+
constexpr bool match_v8qi(__m64 _v, char a, char b, char c, char d, char e, char f, char g, char h) {
__v8qi v = (__v8qi)_v;
return v[0] == a && v[1] == b && v[2] == c && v[3] == d && v[4] == e && v[5] == f && v[6] == g && v[7] == h;
}
+constexpr bool match_v8qu(__m64 _v, unsigned char a, unsigned char b, unsigned char c, unsigned char d, unsigned char e, unsigned char f, unsigned char g, unsigned char h) {
+ __v8qu v = (__v8qu)_v;
+ return v[0] == a && v[1] == b && v[2] == c && v[3] == d && v[4] == e && v[5] == f && v[6] == g && v[7] == h;
+}
+
constexpr bool match_m128(__m128 v, float a, float b, float c, float d) {
return v[0] == a && v[1] == b && v[2] == c && v[3] == d;
}
@@ -60,11 +70,21 @@ constexpr bool match_v8hi(__m128i _v, short a, short b, short c, short d, short
return v[0] == a && v[1] == b && v[2] == c && v[3] == d && v[4] == e && v[5] == f && v[6] == g && v[7] == h;
}
+constexpr bool match_v8hu(__m128i _v, unsigned short a, unsigned short b, unsigned short c, unsigned short d, unsigned short e, unsigned short f, unsigned short g, unsigned short h) {
+ __v8hu v = (__v8hu)_v;
+ return v[0] == a && v[1] == b && v[2] == c && v[3] == d && v[4] == e && v[5] == f && v[6] == g && v[7] == h;
+}
+
constexpr bool match_v16qi(__m128i _v, char a, char b, char c, char d, char e, char f, char g, char h, char i, char j, char k, char l, char m, char n, char o, char p) {
__v16qi v = (__v16qi)_v;
return v[0] == a && v[1] == b && v[2] == c && v[3] == d && v[4] == e && v[5] == f && v[6] == g && v[7] == h && v[8] == i && v[9] == j && v[10] == k && v[11] == l && v[12] == m && v[13] == n && v[14] == o && v[15] == p;
}
+constexpr bool match_v16qu(__m128i _v, unsigned char a, unsigned char b, unsigned char c, unsigned char d, unsigned char e, unsigned char f, unsigned char g, unsigned char h, unsigned char i, unsigned char j, unsigned char k, unsigned char l, unsigned char m, unsigned char n, unsigned char o, unsigned char p) {
+ __v16qu v = (__v16qu)_v;
+ return v[0] == a && v[1] == b && v[2] == c && v[3] == d && v[4] == e && v[5] == f && v[6] == g && v[7] == h && v[8] == i && v[9] == j && v[10] == k && v[11] == l && v[12] == m && v[13] == n && v[14] == o && v[15] == p;
+}
+
constexpr bool match_m256(__m256 v, float a, float b, float c, float d, float e, float f, float g, float h) {
return v[0] == a && v[1] == b && v[2] == c && v[3] == d && v[4] == e && v[5] == f && v[6] == g && v[7] == h;
}
@@ -100,6 +120,11 @@ constexpr bool match_v16hi(__m256i _v, short a, short b, short c, short d, short
return v[0] == a && v[1] == b && v[2] == c && v[3] == d && v[4] == e && v[5] == f && v[6] == g && v[7] == h && v[8] == i && v[9] == j && v[10] == k && v[11] == l && v[12] == m && v[13] == n && v[14] == o && v[15] == p;
}
+constexpr bool match_v16hu(__m256i _v, unsigned short a, unsigned short b, unsigned short c, unsigned short d, unsigned short e, unsigned short f, unsigned short g, unsigned short h, unsigned short i, unsigned short j, unsigned short k, unsigned short l, unsigned short m, unsigned short n, unsigned short o, unsigned short p) {
+ __v16hu v = (__v16hu)_v;
+ return v[0] == a && v[1] == b && v[2] == c && v[3] == d && v[4] == e && v[5] == f && v[6] == g && v[7] == h && v[8] == i && v[9] == j && v[10] == k && v[11] == l && v[12] == m && v[13] == n && v[14] == o && v[15] == p;
+}
+
constexpr bool match_v32qi(__m256i _v, char __b00, char __b01, char __b02, char __b03, char __b04, char __b05, char __b06, char __b07,
char __b08, char __b09, char __b10, char __b11, char __b12, char __b13, char __b14, char __b15,
char __b16, char __b17, char __b18, char __b19, char __b20, char __b21, char __b22, char __b23,
@@ -111,6 +136,17 @@ constexpr bool match_v32qi(__m256i _v, char __b00, char __b01, char __b02, char
v[24] == __b24 && v[25] == __b25 && v[26] == __b26 && v[27] == __b27 && v[28] == __b28 && v[29] == __b29 && v[30] == __b30 && v[31] == __b31;
}
+constexpr bool match_v32qu(__m256i _v, unsigned char __b00, unsigned char __b01, unsigned char __b02, unsigned char __b03, unsigned char __b04, unsigned char __b05, unsigned char __b06, unsigned char __b07,
+ unsigned char __b08, unsigned char __b09, unsigned char __b10, unsigned char __b11, unsigned char __b12, unsigned char __b13, unsigned char __b14, unsigned char __b15,
+ unsigned char __b16, unsigned char __b17, unsigned char __b18, unsigned char __b19, unsigned char __b20, unsigned char __b21, unsigned char __b22, unsigned char __b23,
+ unsigned char __b24, unsigned char __b25, unsigned char __b26, unsigned char __b27, unsigned char __b28, unsigned char __b29, unsigned char __b30, unsigned char __b31) {
+ __v32qu v = (__v32qu)_v;
+ return v[ 0] == __b00 && v[ 1] == __b01 && v[ 2] == __b02 && v[ 3] == __b03 && v[ 4] == __b04 && v[ 5] == __b05 && v[ 6] == __b06 && v[ 7] == __b07 &&
+ v[ 8] == __b08 && v[ 9] == __b09 && v[10] == __b10 && v[11] == __b11 && v[12] == __b12 && v[13] == __b13 && v[14] == __b14 && v[15] == __b15 &&
+ v[16] == __b16 && v[17] == __b17 && v[18] == __b18 && v[19] == __b19 && v[20] == __b20 && v[21] == __b21 && v[22] == __b22 && v[23] == __b23 &&
+ v[24] == __b24 && v[25] == __b25 && v[26] == __b26 && v[27] == __b27 && v[28] == __b28 && v[29] == __b29 && v[30] == __b30 && v[31] == __b31;
+}
+
constexpr bool match_m512(__m512 v, float a, float b, float c, float d, float e, float f, float g, float h, float i, float j, float k, float l, float m, float n, float o, float p) {
return v[0] == a && v[1] == b && v[2] == c && v[3] == d && v[4] == e && v[5] == f && v[6] == g && v[7] == h && v[8] == i && v[9] == j && v[10] == k && v[11] == l && v[12] == m && v[13] == n && v[14] == o && v[15] == p;
}
@@ -156,6 +192,17 @@ constexpr bool match_v32hi(__m512i _v, short __e00, short __e01, short __e02, sh
v[24] == __e24 && v[25] == __e25 && v[26] == __e26 && v[27] == __e27 && v[28] == __e28 && v[29] == __e29 && v[30] == __e30 && v[31] == __e31;
}
+constexpr bool match_v32hu(__m512i _v, unsigned short __e00, unsigned short __e01, unsigned short __e02, unsigned short __e03, unsigned short __e04, unsigned short __e05, unsigned short __e06, unsigned short __e07,
+ unsigned short __e08, unsigned short __e09, unsigned short __e10, unsigned short __e11, unsigned short __e12, unsigned short __e13, unsigned short __e14, unsigned short __e15,
+ unsigned short __e16, unsigned short __e17, unsigned short __e18, unsigned short __e19, unsigned short __e20, unsigned short __e21, unsigned short __e22, unsigned short __e23,
+ unsigned short __e24, unsigned short __e25, unsigned short __e26, unsigned short __e27, unsigned short __e28, unsigned short __e29, unsigned short __e30, unsigned short __e31) {
+ __v32hu v = (__v32hu)_v;
+ return v[ 0] == __e00 && v[ 1] == __e01 && v[ 2] == __e02 && v[ 3] == __e03 && v[ 4] == __e04 && v[ 5] == __e05 && v[ 6] == __e06 && v[ 7] == __e07 &&
+ v[ 8] == __e08 && v[ 9] == __e09 && v[10] == __e10 && v[11] == __e11 && v[12] == __e12 && v[13] == __e13 && v[14] == __e14 && v[15] == __e15 &&
+ v[16] == __e16 && v[17] == __e17 && v[18] == __e18 && v[19] == __e19 && v[20] == __e20 && v[21] == __e21 && v[22] == __e22 && v[23] == __e23 &&
+ v[24] == __e24 && v[25] == __e25 && v[26] == __e26 && v[27] == __e27 && v[28] == __e28 && v[29] == __e29 && v[30] == __e30 && v[31] == __e31;
+}
+
constexpr bool match_v64qi(__m512i _v, char __e00, char __e01, char __e02, char __e03, char __e04, char __e05, char __e06, char __e07,
char __e08, char __e09, char __e10, char __e11, char __e12, char __e13, char __e14, char __e15,
char __e16, char __e17, char __e18, char __e19, char __e20, char __e21, char __e22, char __e23,
@@ -175,6 +222,25 @@ constexpr bool match_v64qi(__m512i _v, char __e00, char __e01, char __e02, char
v[56] == __e56 && v[57] == __e57 && v[58] == __e58 && v[59] == __e59 && v[60] == __e60 && v[61] == __e61 && v[62] == __e62 && v[63] == __e63;
}
+constexpr bool match_v64qu(__m512i _v, unsigned char __e00, unsigned char __e01, unsigned char __e02, unsigned char __e03, unsigned char __e04, unsigned char __e05, unsigned char __e06, unsigned char __e07,
+ unsigned char __e08, unsigned char __e09, unsigned char __e10, unsigned char __e11, unsigned char __e12, unsigned char __e13, unsigned char __e14, unsigned char __e15,
+ unsigned char __e16, unsigned char __e17, unsigned char __e18, unsigned char __e19, unsigned char __e20, unsigned char __e21, unsigned char __e22, unsigned char __e23,
+ unsigned char __e24, unsigned char __e25, unsigned char __e26, unsigned char __e27, unsigned char __e28, unsigned char __e29, unsigned char __e30, unsigned char __e31,
+ unsigned char __e32, unsigned char __e33, unsigned char __e34, unsigned char __e35, unsigned char __e36, unsigned char __e37, unsigned char __e38, unsigned char __e39,
+ unsigned char __e40, unsigned char __e41, unsigned char __e42, unsigned char __e43, unsigned char __e44, unsigned char __e45, unsigned char __e46, unsigned char __e47,
+ unsigned char __e48, unsigned char __e49, unsigned char __e50, unsigned char __e51, unsigned char __e52, unsigned char __e53, unsigned char __e54, unsigned char __e55,
+ unsigned char __e56, unsigned char __e57, unsigned char __e58, unsigned char __e59, unsigned char __e60, unsigned char __e61, unsigned char __e62, unsigned char __e63) {
+ __v64qu v = (__v64qu)_v;
+ return v[ 0] == __e00 && v[ 1] == __e01 && v[ 2] == __e02 && v[ 3] == __e03 && v[ 4] == __e04 && v[ 5] == __e05 && v[ 6] == __e06 && v[ 7] == __e07 &&
+ v[ 8] == __e08 && v[ 9] == __e09 && v[10] == __e10 && v[11] == __e11 && v[12] == __e12 && v[13] == __e13 && v[14] == __e14 && v[15] == __e15 &&
+ v[16] == __e16 && v[17] == __e17 && v[18] == __e18 && v[19] == __e19 && v[20] == __e20 && v[21] == __e21 && v[22] == __e22 && v[23] == __e23 &&
+ v[24] == __e24 && v[25] == __e25 && v[26] == __e26 && v[27] == __e27 && v[28] == __e28 && v[29] == __e29 && v[30] == __e30 && v[31] == __e31 &&
+ v[32] == __e32 && v[33] == __e33 && v[34] == __e34 && v[35] == __e35 && v[36] == __e36 && v[37] == __e37 && v[38] == __e38 && v[39] == __e39 &&
+ v[40] == __e40 && v[41] == __e41 && v[42] == __e42 && v[43] == __e43 && v[44] == __e44 && v[45] == __e45 && v[46] == __e46 && v[47] == __e47 &&
+ v[48] == __e48 && v[49] == __e49 && v[50] == __e50 && v[51] == __e51 && v[52] == __e52 && v[53] == __e53 && v[54] == __e54 && v[55] == __e55 &&
+ v[56] == __e56 && v[57] == __e57 && v[58] == __e58 && v[59] == __e59 && v[60] == __e60 && v[61] == __e61 && v[62] == __e62 && v[63] == __e63;
+}
+
#define TEST_CONSTEXPR(...) static_assert(__VA_ARGS__)
#else
diff --git a/clang/test/CodeGen/X86/mmx-builtins.c b/clang/test/CodeGen/X86/mmx-builtins.c
index 07fab2d39054c..a4098c8db4f3f 100644
--- a/clang/test/CodeGen/X86/mmx-builtins.c
+++ b/clang/test/CodeGen/X86/mmx-builtins.c
@@ -62,24 +62,28 @@ __m64 test_mm_adds_pi8(__m64 a, __m64 b) {
// CHECK: call <8 x i8> @llvm.sadd.sat.v8i8(
return _mm_adds_pi8(a, b);
}
+TEST_CONSTEXPR(match_v8qi(_mm_adds_pi8((__m64)(__v8qs){+100, +50, -100, +20, +80, -50, +120, -20}, (__m64)(__v8qs){+50, +80, -50, +110, +60, -30, +20, -10}), +127, +127, -128, +127, +127, -80, +127, -30));
__m64 test_mm_adds_pi16(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_adds_pi16
// CHECK: call <4 x i16> @llvm.sadd.sat.v4i16(
return _mm_adds_pi16(a, b);
}
+TEST_CONSTEXPR(match_v4hi(_mm_adds_pi16((__m64)(__v4hi){+32000, -32000, +32000, -32000}, (__m64)(__v4hi){+800, -800, -800, +800}), +32767, -32768, +31200, -31200));
__m64 test_mm_adds_pu8(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_adds_pu8
// CHECK: call <8 x i8> @llvm.uadd.sat.v8i8(
return _mm_adds_pu8(a, b);
}
+TEST_CONSTEXPR(match_v8qu(_mm_adds_pu8((__m64)(__v8qu){0, +1, +2, +3, +180, +150, +120, +200}, (__m64)(__v8qu){0, +1, +2, +3, +160, +30, +200, +10}), 0, +2, +4, +6, +255, +180, +255, +210));
__m64 test_mm_adds_pu16(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_adds_pu16
// CHECK: call <4 x i16> @llvm.uadd.sat.v4i16(
return _mm_adds_pu16(a, b);
}
+TEST_CONSTEXPR(match_v4hu(_mm_adds_pu16((__m64)(__v4hu){+0, +1, +32000, +33000}, (__m64)(__v4hu){0, +1, +800, +33000}), 0, +2, +32800, +65535));
__m64 test_mm_alignr_pi8(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_alignr_pi8
@@ -718,24 +722,28 @@ __m64 test_mm_subs_pi8(__m64 a, __m64 b) {
// CHECK: call <8 x i8> @llvm.ssub.sat.v8i8(
return _mm_subs_pi8(a, b);
}
+TEST_CONSTEXPR(match_v8qi(_mm_subs_pi8((__m64)(__v8qs){+100, +50, -100, +20, +80, -50, +120, -20}, (__m64)(__v8qs){-50, -80, +50, -110, -60, +30, -20, +10}), +127, +127, -128, +127, +127, -80, +127, -30));
__m64 test_mm_subs_pi16(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_subs_pi16
// CHECK: call <4 x i16> @llvm.ssub.sat.v4i16(
return _mm_subs_pi16(a, b);
}
+TEST_CONSTEXPR(match_v4hi(_mm_subs_pi16((__m64)(__v4hi){+32000, -32000, +32000, -32000}, (__m64)(__v4hi){-800, +800, +800, -800}), +32767, -32768, +31200, -31200));
__m64 test_mm_subs_pu8(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_subs_pu8
// CHECK: call <8 x i8> @llvm.usub.sat.v8i8(
return _mm_subs_pu8(a, b);
}
+TEST_CONSTEXPR(match_v8qu(_mm_subs_pu8((__m64)(__v8qu){0, +1, +2, +3, +180, +250, +120, +200}, (__m64)(__v8qu){0, +1, +1, +255, +200, +30, +200, +10}), 0, 0, +1, 0, 0, +220, 0, +190));
__m64 test_mm_subs_pu16(__m64 a, __m64 b) {
// CHECK-LABEL: test_mm_subs_pu16
// CHECK: call <4 x i16> @llvm.usub.sat.v4i16(
return _mm_subs_pu16(a, b);
}
+TEST_CONSTEXPR(match_v4hu(_mm_subs_pu16((__m64)(__v4hu){+0, +1, +32000, +33000}, (__m64)(__v4hu){0, +65535, +800, +34000}), 0, 0, +31200, 0));
int test_m_to_int(__m64 a) {
// CHECK-LABEL: test_m_to_int
diff --git a/clang/test/CodeGen/X86/sse2-builtins.c b/clang/test/CodeGen/X86/sse2-builtins.c
index 4ff14adcea0e1..612a619734989 100644
--- a/clang/test/CodeGen/X86/sse2-builtins.c
+++ b/clang/test/CodeGen/X86/sse2-builtins.c
@@ -63,12 +63,14 @@ __m128i test_mm_adds_epi8(__m128i A, __m128i B) {
// CHECK: call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
return _mm_adds_epi8(A, B);
}
+TEST_CONSTEXPR(match_v16qi(_mm_adds_epi8((__m128i)(__v16qs){+100, +50, -100, +20, +80, -50, +120, -20, -100, -50, +100, -20, -80, +50, -120, +20}, (__m128i)(__v16qs){+50, +80, -50, +110, +60, -30, +20, -10, +50, +80, -50, +110, +60, -30, +20, -10}), +127, +127, -128, +127, +127, -80, +127, -30, -50, +30, +50, +90, -20, +20, -100, +10));
__m128i test_mm_adds_epi16(__m128i A, __m128i B) {
// CHECK-LABEL: test_mm_adds_epi16
// CHECK: call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> %{{.*}}, <8 x i16> %{{.*}})
return _mm_adds_epi16(A, B);
}
+TEST_CONSTEXPR(match_v8hi(_mm_adds_epi16((__m128i)(__v8hi){+32000, -32000, +32000, -32000, +80, -50, +120, -20}, (__m128i)(__v8hi){+800, -800, -800, +800, +60, -30, +20, -10}), +32767, -32768, +31200, -31200, +140, -80, +140, -30));
__m128i test_mm_adds_epu8(__m128i A, __m128i B) {
// CHECK-LABEL: test_mm_adds_epu8
@@ -76,6 +78,7 @@ __m128i test_mm_adds_epu8(__m128i A, __m128i B) {
// CHECK: call <16 x i8> @llvm.uadd.sat.v16i8(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
return _mm_adds_epu8(A, B);
}
+TEST_CONSTEXPR(match_v16qu(_mm_adds_epu8((__m128i)(__v16qu){0, 0, 0, 0, +127, +127, +127, +127, +128, +128, +128, +128, +255, +255, +255, +255}, (__m128i)(__v16qu){0, +127, +128, +255, 0, +127, +128, +255, 0, +127, +128, +255, 0, +127, +128, +255}), 0, +127, +128, +255, +127, +254, +255, +255, +128, +255, +255, +255, +255, +255, +255, +255));
__m128i test_mm_adds_epu16(__m128i A, __m128i B) {
// CHECK-LABEL: test_mm_adds_epu16
@@ -83,6 +86,7 @@ __m128i test_mm_adds_epu16(__m128i A, __m128i B) {
// CHECK: call <8 x i16> @llvm.uadd.sat.v8i16(<8 x i16> %{{.*}}, <8 x i16> %{{.*}})
return _mm_adds_epu16(A, B);
}
+TEST_CONSTEXPR(match_v8hu(_mm_adds_epu16((__m128i)(__v8hu){0, 0, 0, +32767, +32767, +32767, +65535, +65535}, (__m128i)(__v8hu){0, +32767, +65535, 0, +32767, +65535, 0, +32767}), 0, +32767, +65535, +32767, +65534, +65535, +65535, +65535));
__m128d test_mm_and_pd(__m128d A, __m128d B) {
// CHECK-LABEL: test_mm_and_pd
@@ -1695,12 +1699,14 @@ __m128i test_mm_subs_epi8(__m128i A, __m128i B) {
// CHECK: call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
return _mm_subs_epi8(A, B);
}
+TEST_CONSTEXPR(match_v16qi(_mm_subs_epi8((__m128i)(__v16qs){+100, +50, -100, +20, +80, -50, +120, -20, -100, -50, +100, -20, -80, +50, -120, +20}, (__m128i)(__v16qs){-50, -80, +50, -110, -60, +30, -20, +10, -50, -80, +50, -110, -60, +30, -20, +10}), +127, +127, -128, +127, +127, -80, +127, -30, -50, +30, +50, +90, -20, +20, -100, +10));
__m128i test_mm_subs_epi16(__m128i A, __m128i B) {
// CHECK-LABEL: test_mm_subs_epi16
// CHECK: call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> %{{.*}}, <8 x i16> %{{.*}})
return _mm_subs_epi16(A, B);
}
+TEST_CONSTEXPR(match_v8hi(_mm_subs_epi16((__m128i)(__v8hi){+32000, -32000, +32000, -32000, +80, -50, +120, -20}, (__m128i)(__v8hi){-800, +800, +800, -800, -60, +30, -20, +10}), +32767, -32768, +31200, -31200, +140, -80, +140, -30));
__m128i test_mm_subs_epu8(__m128i A, __m128i B) {
// CHECK-LABEL: test_mm_subs_epu8
@@ -1708,6 +1714,7 @@ __m128i test_mm_subs_epu8(__m128i A, __m128i B) {
// CHECK: call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> %{{.*}}, <16 x i8> %{{.*}})
return _mm_subs_epu8(A, B);
}
+TEST_CONSTEXPR(match_v16qu(_mm_subs_epu8((__m128i)(__v16qu){0, 0, 0, 0, +127, +127, +127, +127, +128, +128, +128, +128, +255, +255, +255, +255}, (__m128i)(__v16qu){0, +127, +128, +255, 0, +127, +128, +255, 0, +127, +128, +255, 0, +127, +128, +255}), 0, 0, 0, 0, +127, 0, 0, 0, +128, +1, 0, 0, +255, +128, +127, 0));
__m128i test_mm_subs_epu16(__m128i A, __m128i B) {
// CHECK-LABEL: test_mm_subs_epu16
@@ -1715,6 +1722,7 @@ __m128i test_mm_subs_epu16(__m128i A, __m128i B) {
// CHECK: call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> %{{.*}}, <8 x i16> %{{.*}})
return _mm_subs_epu16(A, B);
}
+TEST_CONSTEXPR(match_v8hu(_mm_subs_epu16((__m128i)(__v8hu){0, 0, 0, +32767, +32767, +32767, +65535, +65535}, (__m128i)(__v8hu){0, +32767, +65535, 0, +32767, +65535, 0, +32767}), 0, 0, 0, +32767, 0, 0, +65535, +32768));
int test_mm_ucomieq_sd(__m128d A, __m128d B) {
// CHECK-LABEL: test_mm_ucomieq_sd
More information about the cfe-commits
mailing list