https://github.com/ckoparkar updated
https://github.com/llvm/llvm-project/pull/155199
>From 981fa7db5cd83b9c8b240e6b116a52592a59617a Mon Sep 17 00:00:00 2001
From: Chaitanya Koparkar <ckopar...@gmail.com>
Date: Sat, 23 Aug 2025 07:57:38 -0400
Subject: [PATCH] [Headers][X86] Allow AVX512VLBW integer reduction intrinsics
to be used in constexpr
Add constexpr support for the following:
_mm_reduce_add_epi8 _mm_reduce_add_epi16 _mm256_reduce_add_epi8
_mm256_reduce_add_epi16
_mm_reduce_mul_epi8 _mm_reduce_mul_epi16 _mm256_reduce_mul_epi8
_mm256_reduce_mul_epi16
_mm_reduce_and_epi8 _mm_reduce_and_epi16 _mm256_reduce_and_epi8
_mm256_reduce_and_epi16
_mm_reduce_or_epi8 _mm_reduce_or_epi16 _mm256_reduce_or_epi8
_mm256_reduce_or_epi16
_mm_mask_reduce_add_epi8 _mm_mask_reduce_add_epi16 _mm256_mask_reduce_add_epi8
_mm256_mask_reduce_add_epi16
_mm_mask_reduce_mul_epi8 _mm_mask_reduce_mul_epi16 _mm256_mask_reduce_mul_epi8
_mm256_mask_reduce_mul_epi16
_mm_mask_reduce_and_epi8 _mm_mask_reduce_and_epi16 _mm256_mask_reduce_and_epi8
_mm256_mask_reduce_and_epi16
_mm_mask_reduce_or_epi8 _mm_mask_reduce_or_epi16 _mm256_mask_reduce_or_epi8
_mm256_mask_reduce_or_epi16
_mm_reduce_max_epi8 _mm_reduce_max_epi16 _mm256_reduce_max_epi8
_mm256_reduce_max_epi16
_mm_reduce_min_epi8 _mm_reduce_min_epi16 _mm256_reduce_min_epi8
_mm256_reduce_min_epi16
_mm_reduce_max_epu8 _mm_reduce_max_epu16 _mm256_reduce_max_epu8
_mm256_reduce_max_epu16
_mm_reduce_min_epu8 _mm_reduce_min_epu16 _mm256_reduce_min_epu8
_mm256_reduce_min_epu16
_mm_mask_reduce_max_epi8 _mm_mask_reduce_max_epi16 _mm256_mask_reduce_max_epi8
_mm256_mask_reduce_max_epi16
_mm_mask_reduce_min_epi8 _mm_mask_reduce_min_epi16 _mm256_mask_reduce_min_epi8
_mm256_mask_reduce_min_epi16
_mm_mask_reduce_max_epu8 _mm_mask_reduce_max_epu16 _mm256_mask_reduce_max_epu8
_mm256_mask_reduce_max_epu16
_mm_mask_reduce_min_epu8 _mm_mask_reduce_min_epu16 _mm256_mask_reduce_min_epu8
_mm256_mask_reduce_min_epu16
---
clang/lib/Headers/avx512vlbwintrin.h | 183 +++++++++---------
.../CodeGen/X86/avx512vlbw-reduceIntrin.c | 97 ++++++++++
2 files changed, 184 insertions(+), 96 deletions(-)
diff --git a/clang/lib/Headers/avx512vlbwintrin.h
b/clang/lib/Headers/avx512vlbwintrin.h
index 2e2052ad1b682..ec1170c28039b 100644
--- a/clang/lib/Headers/avx512vlbwintrin.h
+++ b/clang/lib/Headers/avx512vlbwintrin.h
@@ -2197,71 +2197,62 @@ _mm256_maskz_srli_epi16(__mmask16 __U, __m256i __A, int
__B)
(__v16hi)_mm256_setzero_si256());
}
-static __inline__ __m128i __DEFAULT_FN_ATTRS128
-_mm_mask_mov_epi16 (__m128i __W, __mmask8 __U, __m128i __A)
-{
+static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask_mov_epi16(__m128i __W, __mmask8 __U, __m128i __A) {
return (__m128i) __builtin_ia32_selectw_128 ((__mmask8) __U,
(__v8hi) __A,
(__v8hi) __W);
}
-static __inline__ __m128i __DEFAULT_FN_ATTRS128
-_mm_maskz_mov_epi16 (__mmask8 __U, __m128i __A)
-{
+static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_maskz_mov_epi16(__mmask8 __U, __m128i __A) {
return (__m128i) __builtin_ia32_selectw_128 ((__mmask8) __U,
(__v8hi) __A,
(__v8hi) _mm_setzero_si128 ());
}
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_mask_mov_epi16 (__m256i __W, __mmask16 __U, __m256i __A)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask_mov_epi16(__m256i __W, __mmask16 __U, __m256i __A) {
return (__m256i) __builtin_ia32_selectw_256 ((__mmask16) __U,
(__v16hi) __A,
(__v16hi) __W);
}
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_maskz_mov_epi16 (__mmask16 __U, __m256i __A)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_maskz_mov_epi16(__mmask16 __U, __m256i __A) {
return (__m256i) __builtin_ia32_selectw_256 ((__mmask16) __U,
(__v16hi) __A,
(__v16hi) _mm256_setzero_si256 ());
}
-static __inline__ __m128i __DEFAULT_FN_ATTRS128
-_mm_mask_mov_epi8 (__m128i __W, __mmask16 __U, __m128i __A)
-{
+static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask_mov_epi8(__m128i __W, __mmask16 __U, __m128i __A) {
return (__m128i) __builtin_ia32_selectb_128 ((__mmask16) __U,
(__v16qi) __A,
(__v16qi) __W);
}
-static __inline__ __m128i __DEFAULT_FN_ATTRS128
-_mm_maskz_mov_epi8 (__mmask16 __U, __m128i __A)
-{
+static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_maskz_mov_epi8(__mmask16 __U, __m128i __A) {
return (__m128i) __builtin_ia32_selectb_128 ((__mmask16) __U,
(__v16qi) __A,
(__v16qi) _mm_setzero_si128 ());
}
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_mask_mov_epi8 (__m256i __W, __mmask32 __U, __m256i __A)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask_mov_epi8(__m256i __W, __mmask32 __U, __m256i __A) {
return (__m256i) __builtin_ia32_selectb_256 ((__mmask32) __U,
(__v32qi) __A,
(__v32qi) __W);
}
-static __inline__ __m256i __DEFAULT_FN_ATTRS256
-_mm256_maskz_mov_epi8 (__mmask32 __U, __m256i __A)
-{
+static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_maskz_mov_epi8(__mmask32 __U, __m256i __A) {
return (__m256i) __builtin_ia32_selectb_256 ((__mmask32) __U,
(__v32qi) __A,
(__v32qi) _mm256_setzero_si256 ());
}
-
static __inline__ __m128i __DEFAULT_FN_ATTRS128
_mm_mask_set1_epi8 (__m128i __O, __mmask16 __M, char __A)
{
@@ -2812,353 +2803,353 @@ _mm256_mask_permutexvar_epi16 (__m256i __W, __mmask16
__M, __m256i __A,
(__v16hi)_mm256_dbsad_epu8((A), (B), (imm)),
\
(__v16hi)_mm256_setzero_si256()))
-static __inline__ short __DEFAULT_FN_ATTRS128
-_mm_reduce_add_epi16(__m128i __W) {
+static __inline__ short
+ __DEFAULT_FN_ATTRS128_CONSTEXPR _mm_reduce_add_epi16(__m128i __W) {
return __builtin_reduce_add((__v8hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS128
+static __inline__ short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_mul_epi16(__m128i __W) {
return __builtin_reduce_mul((__v8hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS128
+static __inline__ short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_and_epi16(__m128i __W) {
return __builtin_reduce_and((__v8hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS128
+static __inline__ short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_or_epi16(__m128i __W) {
return __builtin_reduce_or((__v8hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS128
-_mm_mask_reduce_add_epi16( __mmask8 __M, __m128i __W) {
+static __inline__ short __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask_reduce_add_epi16(__mmask8 __M, __m128i __W) {
__W = _mm_maskz_mov_epi16(__M, __W);
return __builtin_reduce_add((__v8hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS128
-_mm_mask_reduce_mul_epi16( __mmask8 __M, __m128i __W) {
+static __inline__ short __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask_reduce_mul_epi16(__mmask8 __M, __m128i __W) {
__W = _mm_mask_mov_epi16(_mm_set1_epi16(1), __M, __W);
return __builtin_reduce_mul((__v8hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS128
-_mm_mask_reduce_and_epi16( __mmask8 __M, __m128i __W) {
+static __inline__ short __DEFAULT_FN_ATTRS128_CONSTEXPR
+_mm_mask_reduce_and_epi16(__mmask8 __M, __m128i __W) {
__W = _mm_mask_mov_epi16(_mm_set1_epi16(-1), __M, __W);
return __builtin_reduce_and((__v8hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS128
+static __inline__ short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_or_epi16(__mmask8 __M, __m128i __W) {
__W = _mm_maskz_mov_epi16(__M, __W);
return __builtin_reduce_or((__v8hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS128
+static __inline__ short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_max_epi16(__m128i __V) {
return __builtin_reduce_max((__v8hi)__V);
}
-static __inline__ unsigned short __DEFAULT_FN_ATTRS128
+static __inline__ unsigned short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_max_epu16(__m128i __V) {
return __builtin_reduce_max((__v8hu)__V);
}
-static __inline__ short __DEFAULT_FN_ATTRS128
+static __inline__ short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_min_epi16(__m128i __V) {
return __builtin_reduce_min((__v8hi)__V);
}
-static __inline__ unsigned short __DEFAULT_FN_ATTRS128
+static __inline__ unsigned short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_min_epu16(__m128i __V) {
return __builtin_reduce_min((__v8hu)__V);
}
-static __inline__ short __DEFAULT_FN_ATTRS128
+static __inline__ short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_max_epi16(__mmask16 __M, __m128i __V) {
__V = _mm_mask_mov_epi16(_mm_set1_epi16(-32767-1), __M, __V);
return __builtin_reduce_max((__v8hi)__V);
}
-static __inline__ unsigned short __DEFAULT_FN_ATTRS128
+static __inline__ unsigned short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_max_epu16(__mmask16 __M, __m128i __V) {
__V = _mm_maskz_mov_epi16(__M, __V);
return __builtin_reduce_max((__v8hu)__V);
}
-static __inline__ short __DEFAULT_FN_ATTRS128
+static __inline__ short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_min_epi16(__mmask16 __M, __m128i __V) {
__V = _mm_mask_mov_epi16(_mm_set1_epi16(32767), __M, __V);
return __builtin_reduce_min((__v8hi)__V);
}
-static __inline__ unsigned short __DEFAULT_FN_ATTRS128
+static __inline__ unsigned short __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_min_epu16(__mmask16 __M, __m128i __V) {
__V = _mm_mask_mov_epi16(_mm_set1_epi16(-1), __M, __V);
return __builtin_reduce_min((__v8hu)__V);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_add_epi16(__m256i __W) {
return __builtin_reduce_add((__v16hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_mul_epi16(__m256i __W) {
return __builtin_reduce_mul((__v16hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_and_epi16(__m256i __W) {
return __builtin_reduce_and((__v16hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_or_epi16(__m256i __W) {
return __builtin_reduce_or((__v16hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
-_mm256_mask_reduce_add_epi16( __mmask16 __M, __m256i __W) {
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask_reduce_add_epi16(__mmask16 __M, __m256i __W) {
__W = _mm256_maskz_mov_epi16(__M, __W);
return __builtin_reduce_add((__v16hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
-_mm256_mask_reduce_mul_epi16( __mmask16 __M, __m256i __W) {
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask_reduce_mul_epi16(__mmask16 __M, __m256i __W) {
__W = _mm256_mask_mov_epi16(_mm256_set1_epi16(1), __M, __W);
return __builtin_reduce_mul((__v16hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
-_mm256_mask_reduce_and_epi16( __mmask16 __M, __m256i __W) {
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
+_mm256_mask_reduce_and_epi16(__mmask16 __M, __m256i __W) {
__W = _mm256_mask_mov_epi16(_mm256_set1_epi16(-1), __M, __W);
return __builtin_reduce_and((__v16hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_or_epi16(__mmask16 __M, __m256i __W) {
__W = _mm256_maskz_mov_epi16(__M, __W);
return __builtin_reduce_or((__v16hi)__W);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_max_epi16(__m256i __V) {
return __builtin_reduce_max((__v16hi)__V);
}
-static __inline__ unsigned short __DEFAULT_FN_ATTRS256
+static __inline__ unsigned short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_max_epu16(__m256i __V) {
return __builtin_reduce_max((__v16hu)__V);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_min_epi16(__m256i __V) {
return __builtin_reduce_min((__v16hi)__V);
}
-static __inline__ unsigned short __DEFAULT_FN_ATTRS256
+static __inline__ unsigned short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_min_epu16(__m256i __V) {
return __builtin_reduce_min((__v16hu)__V);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_max_epi16(__mmask16 __M, __m256i __V) {
__V = _mm256_mask_mov_epi16(_mm256_set1_epi16(-32767-1), __M, __V);
return __builtin_reduce_max((__v16hi)__V);
}
-static __inline__ unsigned short __DEFAULT_FN_ATTRS256
+static __inline__ unsigned short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_max_epu16(__mmask16 __M, __m256i __V) {
__V = _mm256_maskz_mov_epi16(__M, __V);
return __builtin_reduce_max((__v16hu)__V);
}
-static __inline__ short __DEFAULT_FN_ATTRS256
+static __inline__ short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_min_epi16(__mmask16 __M, __m256i __V) {
__V = _mm256_mask_mov_epi16(_mm256_set1_epi16(32767), __M, __V);
return __builtin_reduce_min((__v16hi)__V);
}
-static __inline__ unsigned short __DEFAULT_FN_ATTRS256
+static __inline__ unsigned short __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_min_epu16(__mmask16 __M, __m256i __V) {
__V = _mm256_mask_mov_epi16(_mm256_set1_epi16(-1), __M, __V);
return __builtin_reduce_min((__v16hu)__V);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_add_epi8(__m128i __W) {
return __builtin_reduce_add((__v16qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_mul_epi8(__m128i __W) {
return __builtin_reduce_mul((__v16qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_and_epi8(__m128i __W) {
return __builtin_reduce_and((__v16qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_or_epi8(__m128i __W) {
return __builtin_reduce_or((__v16qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_add_epi8(__mmask16 __M, __m128i __W) {
__W = _mm_maskz_mov_epi8(__M, __W);
return __builtin_reduce_add((__v16qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_mul_epi8(__mmask16 __M, __m128i __W) {
__W = _mm_mask_mov_epi8(_mm_set1_epi8(1), __M, __W);
return __builtin_reduce_mul((__v16qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_and_epi8(__mmask16 __M, __m128i __W) {
__W = _mm_mask_mov_epi8(_mm_set1_epi8(-1), __M, __W);
return __builtin_reduce_and((__v16qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_or_epi8(__mmask16 __M, __m128i __W) {
__W = _mm_maskz_mov_epi8(__M, __W);
return __builtin_reduce_or((__v16qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_max_epi8(__m128i __V) {
return __builtin_reduce_max((__v16qs)__V);
}
-static __inline__ unsigned char __DEFAULT_FN_ATTRS128
+static __inline__ unsigned char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_max_epu8(__m128i __V) {
return __builtin_reduce_max((__v16qu)__V);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_min_epi8(__m128i __V) {
return __builtin_reduce_min((__v16qs)__V);
}
-static __inline__ unsigned char __DEFAULT_FN_ATTRS128
+static __inline__ unsigned char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_reduce_min_epu8(__m128i __V) {
return __builtin_reduce_min((__v16qu)__V);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_max_epi8(__mmask16 __M, __m128i __V) {
__V = _mm_mask_mov_epi8(_mm_set1_epi8(-127-1), __M, __V);
return __builtin_reduce_max((__v16qs)__V);
}
-static __inline__ unsigned char __DEFAULT_FN_ATTRS128
+static __inline__ unsigned char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_max_epu8(__mmask16 __M, __m128i __V) {
__V = _mm_maskz_mov_epi8(__M, __V);
return __builtin_reduce_max((__v16qu)__V);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS128
+static __inline__ signed char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_min_epi8(__mmask16 __M, __m128i __V) {
__V = _mm_mask_mov_epi8(_mm_set1_epi8(127), __M, __V);
return __builtin_reduce_min((__v16qs)__V);
}
-static __inline__ unsigned char __DEFAULT_FN_ATTRS128
+static __inline__ unsigned char __DEFAULT_FN_ATTRS128_CONSTEXPR
_mm_mask_reduce_min_epu8(__mmask16 __M, __m128i __V) {
__V = _mm_mask_mov_epi8(_mm_set1_epi8(-1), __M, __V);
return __builtin_reduce_min((__v16qu)__V);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_add_epi8(__m256i __W) {
return __builtin_reduce_add((__v32qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_mul_epi8(__m256i __W) {
return __builtin_reduce_mul((__v32qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_and_epi8(__m256i __W) {
return __builtin_reduce_and((__v32qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_or_epi8(__m256i __W) {
return __builtin_reduce_or((__v32qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_add_epi8(__mmask32 __M, __m256i __W) {
__W = _mm256_maskz_mov_epi8(__M, __W);
return __builtin_reduce_add((__v32qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_mul_epi8(__mmask32 __M, __m256i __W) {
__W = _mm256_mask_mov_epi8(_mm256_set1_epi8(1), __M, __W);
return __builtin_reduce_mul((__v32qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_and_epi8(__mmask32 __M, __m256i __W) {
__W = _mm256_mask_mov_epi8(_mm256_set1_epi8(-1), __M, __W);
return __builtin_reduce_and((__v32qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_or_epi8(__mmask32 __M, __m256i __W) {
__W = _mm256_maskz_mov_epi8(__M, __W);
return __builtin_reduce_or((__v32qs)__W);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_max_epi8(__m256i __V) {
return __builtin_reduce_max((__v32qs)__V);
}
-static __inline__ unsigned char __DEFAULT_FN_ATTRS256
+static __inline__ unsigned char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_max_epu8(__m256i __V) {
return __builtin_reduce_max((__v32qu)__V);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_min_epi8(__m256i __V) {
return __builtin_reduce_min((__v32qs)__V);
}
-static __inline__ unsigned char __DEFAULT_FN_ATTRS256
+static __inline__ unsigned char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_reduce_min_epu8(__m256i __V) {
return __builtin_reduce_min((__v32qu)__V);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_max_epi8(__mmask32 __M, __m256i __V) {
__V = _mm256_mask_mov_epi8(_mm256_set1_epi8(-127-1), __M, __V);
return __builtin_reduce_max((__v32qs)__V);
}
-static __inline__ unsigned char __DEFAULT_FN_ATTRS256
+static __inline__ unsigned char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_max_epu8(__mmask32 __M, __m256i __V) {
__V = _mm256_maskz_mov_epi8(__M, __V);
return __builtin_reduce_max((__v32qu)__V);
}
-static __inline__ signed char __DEFAULT_FN_ATTRS256
+static __inline__ signed char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_min_epi8(__mmask32 __M, __m256i __V) {
__V = _mm256_mask_mov_epi8(_mm256_set1_epi8(127), __M, __V);
return __builtin_reduce_min((__v32qs)__V);
}
-static __inline__ unsigned char __DEFAULT_FN_ATTRS256
+static __inline__ unsigned char __DEFAULT_FN_ATTRS256_CONSTEXPR
_mm256_mask_reduce_min_epu8(__mmask32 __M, __m256i __V) {
__V = _mm256_mask_mov_epi8(_mm256_set1_epi8(-1), __M, __V);
return __builtin_reduce_min((__v32qu)__V);
diff --git a/clang/test/CodeGen/X86/avx512vlbw-reduceIntrin.c
b/clang/test/CodeGen/X86/avx512vlbw-reduceIntrin.c
index faa3b54624a77..8e53db3f45300 100644
--- a/clang/test/CodeGen/X86/avx512vlbw-reduceIntrin.c
+++ b/clang/test/CodeGen/X86/avx512vlbw-reduceIntrin.c
@@ -8,30 +8,35 @@
// RUN: %clang_cc1 -x c++ -ffreestanding %s -O0 -triple=i386 -target-feature
+avx512bw -target-feature +avx512vl -fno-signed-char -emit-llvm -o - -Wall
-Werror | FileCheck %s
#include <immintrin.h>
+#include "builtin_test_helpers.h"
short test_mm_reduce_add_epi16(__m128i __W){
// CHECK-LABEL: test_mm_reduce_add_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %{{.*}})
return _mm_reduce_add_epi16(__W);
}
+TEST_CONSTEXPR(_mm_reduce_add_epi16((__m128i)(__v8hi){1,2,3,4,5,6,7,8}) == 36);
short test_mm_reduce_mul_epi16(__m128i __W){
// CHECK-LABEL: test_mm_reduce_mul_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.mul.v8i16(<8 x i16> %{{.*}})
return _mm_reduce_mul_epi16(__W);
}
+TEST_CONSTEXPR(_mm_reduce_mul_epi16((__m128i)(__v8hi){1,2,3,1,2,3,1,2}) == 72);
short test_mm_reduce_or_epi16(__m128i __W){
// CHECK-LABEL: test_mm_reduce_or_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.or.v8i16(<8 x i16> %{{.*}})
return _mm_reduce_or_epi16(__W);
}
+TEST_CONSTEXPR(_mm_reduce_or_epi16((__m128i)(__v8hi){1,2,4,8,0,0,0,0}) == 15);
short test_mm_reduce_and_epi16(__m128i __W){
// CHECK-LABEL: test_mm_reduce_and_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.and.v8i16(<8 x i16> %{{.*}})
return _mm_reduce_and_epi16(__W);
}
+TEST_CONSTEXPR(_mm_reduce_and_epi16((__m128i)(__v8hi){1,3,5,7,9,11,13,15}) ==
1);
short test_mm_mask_reduce_add_epi16(__mmask8 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_add_epi16
@@ -39,6 +44,8 @@ short test_mm_mask_reduce_add_epi16(__mmask8 __M, __m128i
__W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %{{.*}})
return _mm_mask_reduce_add_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_add_epi16((__mmask8)0b11110000,
(__m128i)(__v8hi){1,2,3,4,5,6,7,8}) == 26);
+TEST_CONSTEXPR(_mm_mask_reduce_add_epi16((__mmask8)0b00001111,
(__m128i)(__v8hi){1,2,3,4,5,6,7,8}) == 10);
short test_mm_mask_reduce_mul_epi16(__mmask8 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_mul_epi16
@@ -46,6 +53,8 @@ short test_mm_mask_reduce_mul_epi16(__mmask8 __M, __m128i
__W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.mul.v8i16(<8 x i16> %{{.*}})
return _mm_mask_reduce_mul_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_mul_epi16((__mmask8)0b11110000,
(__m128i)(__v8hi){1,2,3,1,2,3,1,2}) == 12);
+TEST_CONSTEXPR(_mm_mask_reduce_mul_epi16((__mmask8)0b00001111,
(__m128i)(__v8hi){1,2,3,1,2,3,1,2}) == 6);
short test_mm_mask_reduce_and_epi16(__mmask8 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_and_epi16
@@ -53,6 +62,8 @@ short test_mm_mask_reduce_and_epi16(__mmask8 __M, __m128i
__W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.and.v8i16(<8 x i16> %{{.*}}
return _mm_mask_reduce_and_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_and_epi16((__mmask8)0b11110000,
(__m128i)(__v8hi){1,3,5,7,0,0,0,0}) == 0);
+TEST_CONSTEXPR(_mm_mask_reduce_and_epi16((__mmask8)0b00001111,
(__m128i)(__v8hi){1,3,5,7,0,0,0,0}) == 1);
short test_mm_mask_reduce_or_epi16(__mmask8 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_or_epi16
@@ -60,30 +71,36 @@ short test_mm_mask_reduce_or_epi16(__mmask8 __M, __m128i
__W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.or.v8i16(<8 x i16> %{{.*}})
return _mm_mask_reduce_or_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_or_epi16((__mmask8)0b11110000,
(__m128i)(__v8hi){1,2,4,8,0,0,0,0}) == 0);
+TEST_CONSTEXPR(_mm_mask_reduce_or_epi16((__mmask8)0b00001111,
(__m128i)(__v8hi){1,2,4,8,0,0,0,0}) == 15);
short test_mm256_reduce_add_epi16(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_add_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %{{.*}})
return _mm256_reduce_add_epi16(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_add_epi16((__m256i)(__v16hi){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15})
== 120);
short test_mm256_reduce_mul_epi16(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_mul_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.mul.v16i16(<16 x i16> %{{.*}})
return _mm256_reduce_mul_epi16(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_mul_epi16((__m256i)(__v16hi){1,2,3,1,2,3,1,2,3,1,2,3,1,2,3,1})
== 7776);
short test_mm256_reduce_or_epi16(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_or_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.or.v16i16(<16 x i16> %{{.*}})
return _mm256_reduce_or_epi16(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_or_epi16((__m256i)(__v16hi){1,2,4,8,16,32,64,128,0,0,0,0,0,0,0,0})
== 255);
short test_mm256_reduce_and_epi16(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_and_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.and.v16i16(<16 x i16> %{{.*}})
return _mm256_reduce_and_epi16(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_and_epi16((__m256i)(__v16hi){1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31})
== 1);
short test_mm256_mask_reduce_add_epi16(__mmask16 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_add_epi16
@@ -91,6 +108,8 @@ short test_mm256_mask_reduce_add_epi16(__mmask16 __M,
__m256i __W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %{{.*}})
return _mm256_mask_reduce_add_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_add_epi16((__mmask16)0b1111111100000000,
(__m256i)(__v16hi){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15}) == 84);
+TEST_CONSTEXPR(_mm256_mask_reduce_add_epi16((__mmask16)0b0000000011111111,
(__m256i)(__v16hi){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15}) == 36);
short test_mm256_mask_reduce_mul_epi16(__mmask16 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_mul_epi16
@@ -98,6 +117,8 @@ short test_mm256_mask_reduce_mul_epi16(__mmask16 __M,
__m256i __W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.mul.v16i16(<16 x i16> %{{.*}})
return _mm256_mask_reduce_mul_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_mul_epi16((__mmask16)0b1111111100000000,
(__m256i)(__v16hi){1,2,3,1,2,3,1,2,3,1,2,3,1,2,3,1}) == 108);
+TEST_CONSTEXPR(_mm256_mask_reduce_mul_epi16((__mmask16)0b0000000011111111,
(__m256i)(__v16hi){1,2,3,1,2,3,1,2,3,1,2,3,1,2,3,1}) == 72);
short test_mm256_mask_reduce_and_epi16(__mmask16 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_and_epi16
@@ -105,6 +126,8 @@ short test_mm256_mask_reduce_and_epi16(__mmask16 __M,
__m256i __W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.and.v16i16(<16 x i16> %{{.*}})
return _mm256_mask_reduce_and_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_and_epi16((__mmask16)0b1111111100000000,
(__m256i)(__v16hi){1,3,5,7,9,11,13,15,0,0,0,0,0,0,0,0}) == 0);
+TEST_CONSTEXPR(_mm256_mask_reduce_and_epi16((__mmask16)0b0000000011111111,
(__m256i)(__v16hi){1,3,5,7,9,11,13,15,0,0,0,0,0,0,0,0}) == 1);
short test_mm256_mask_reduce_or_epi16(__mmask16 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_or_epi16
@@ -112,30 +135,36 @@ short test_mm256_mask_reduce_or_epi16(__mmask16 __M,
__m256i __W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.or.v16i16(<16 x i16> %{{.*}})
return _mm256_mask_reduce_or_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_or_epi16((__mmask16)0b1111111100000000,
(__m256i)(__v16hi){1,2,4,8,16,32,64,128,0,0,0,0,0,0,0,0}) == 0);
+TEST_CONSTEXPR(_mm256_mask_reduce_or_epi16((__mmask16)0b0000000011111111,
(__m256i)(__v16hi){1,2,4,8,16,32,64,128,0,0,0,0,0,0,0,0}) == 255);
signed char test_mm_reduce_add_epi8(__m128i __W){
// CHECK-LABEL: test_mm_reduce_add_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.add.v16i8(<16 x i8> %{{.*}})
return _mm_reduce_add_epi8(__W);
}
+TEST_CONSTEXPR(_mm_reduce_add_epi8((__m128i)(__v16qs){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15})
== 120);
signed char test_mm_reduce_mul_epi8(__m128i __W){
// CHECK-LABEL: test_mm_reduce_mul_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.mul.v16i8(<16 x i8> %{{.*}})
return _mm_reduce_mul_epi8(__W);
}
+TEST_CONSTEXPR(_mm_reduce_mul_epi8((__m128i)(__v16qs){1,2,1,1,2,1,1,2,1,1,2,1,1,2,1,1})
== 32);
signed char test_mm_reduce_and_epi8(__m128i __W){
// CHECK-LABEL: test_mm_reduce_and_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.and.v16i8(<16 x i8> %{{.*}})
return _mm_reduce_and_epi8(__W);
}
+TEST_CONSTEXPR(_mm_reduce_and_epi8((__m128i)(__v16qs){1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31})
== 1);
signed char test_mm_reduce_or_epi8(__m128i __W){
// CHECK-LABEL: test_mm_reduce_or_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.or.v16i8(<16 x i8> %{{.*}})
return _mm_reduce_or_epi8(__W);
}
+TEST_CONSTEXPR(_mm_reduce_or_epi8((__m128i)(__v16qs){0,1,2,4,8,16,32,64,0,0,0,0,0,0,0,0})
== 127);
signed char test_mm_mask_reduce_add_epi8(__mmask16 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_add_epi8
@@ -143,6 +172,8 @@ signed char test_mm_mask_reduce_add_epi8(__mmask16 __M,
__m128i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.add.v16i8(<16 x i8> %{{.*}})
return _mm_mask_reduce_add_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_add_epi8((__mmask16)0b1111111100000000,
(__m128i)(__v16qs){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15}) == 84);
+TEST_CONSTEXPR(_mm_mask_reduce_add_epi8((__mmask16)0b0000000011111111,
(__m128i)(__v16qs){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15}) == 36);
signed char test_mm_mask_reduce_mul_epi8(__mmask16 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_mul_epi8
@@ -150,6 +181,8 @@ signed char test_mm_mask_reduce_mul_epi8(__mmask16 __M,
__m128i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.mul.v16i8(<16 x i8> %{{.*}})
return _mm_mask_reduce_mul_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_mul_epi8((__mmask16)0b1111111100000000,
(__m128i)(__v16qs){1,2,1,1,2,1,1,2,1,1,2,1,1,2,1,1}) == 4);
+TEST_CONSTEXPR(_mm_mask_reduce_mul_epi8((__mmask16)0b0000000011111111,
(__m128i)(__v16qs){1,2,1,1,2,1,1,2,1,1,2,1,1,2,1,1}) == 8);
signed char test_mm_mask_reduce_and_epi8(__mmask16 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_and_epi8
@@ -157,6 +190,8 @@ signed char test_mm_mask_reduce_and_epi8(__mmask16 __M,
__m128i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.and.v16i8(<16 x i8> %{{.*}})
return _mm_mask_reduce_and_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_and_epi8((__mmask16)0b1111111100000000,
(__m128i)(__v16qs){1,3,5,7,9,11,13,15,0,0,0,0,0,0,0,0}) == 0);
+TEST_CONSTEXPR(_mm_mask_reduce_and_epi8((__mmask16)0b0000000011111111,
(__m128i)(__v16qs){1,3,5,7,9,11,13,15,0,0,0,0,0,0,0,0}) == 1);
signed char test_mm_mask_reduce_or_epi8(__mmask16 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_or_epi8
@@ -164,30 +199,36 @@ signed char test_mm_mask_reduce_or_epi8(__mmask16 __M,
__m128i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.or.v16i8(<16 x i8> %{{.*}})
return _mm_mask_reduce_or_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_or_epi8((__mmask16)0b1111111100000000,
(__m128i)(__v16qs){0,1,2,4,8,16,32,64,0,0,0,0,0,0,0,0}) == 0);
+TEST_CONSTEXPR(_mm_mask_reduce_or_epi8((__mmask16)0b0000000011111111,
(__m128i)(__v16qs){0,1,2,4,8,16,32,64,0,0,0,0,0,0,0,0}) == 127);
signed char test_mm256_reduce_add_epi8(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_add_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.add.v32i8(<32 x i8> %{{.*}})
return _mm256_reduce_add_epi8(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_add_epi8((__m256i)(__v32qs){0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7})
== 112);
signed char test_mm256_reduce_mul_epi8(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_mul_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.mul.v32i8(<32 x i8> %{{.*}})
return _mm256_reduce_mul_epi8(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_mul_epi8((__m256i)(__v32qs){1,1,1,1,1,1,1,2,1,1,1,1,1,1,1,2,1,1,1,1,1,1,1,2,1,1,1,1,1,1,1,2})
== 16);
signed char test_mm256_reduce_and_epi8(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_and_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.and.v32i8(<32 x i8> %{{.*}})
return _mm256_reduce_and_epi8(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_and_epi8((__m256i)(__v32qs){1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63})
== 1);
signed char test_mm256_reduce_or_epi8(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_or_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.or.v32i8(<32 x i8> %{{.*}})
return _mm256_reduce_or_epi8(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_or_epi8((__m256i)(__v32qs){1,2,4,8,16,32,64,127,1,2,4,8,16,32,64,127,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0})
== 127);
signed char test_mm256_mask_reduce_add_epi8(__mmask32 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_add_epi8
@@ -195,6 +236,8 @@ signed char test_mm256_mask_reduce_add_epi8(__mmask32 __M,
__m256i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.add.v32i8(<32 x i8> %{{.*}})
return _mm256_mask_reduce_add_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_add_epi8((__mmask32)0b11111111111111110000000000000000,
(__m256i)(__v32qs){0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7})
== 56);
+TEST_CONSTEXPR(_mm256_mask_reduce_add_epi8((__mmask32)0b00000000000000001111111111111111,
(__m256i)(__v32qs){8,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7})
== 64);
signed char test_mm256_mask_reduce_mul_epi8(__mmask32 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_mul_epi8
@@ -202,6 +245,8 @@ signed char test_mm256_mask_reduce_mul_epi8(__mmask32 __M,
__m256i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.mul.v32i8(<32 x i8> %{{.*}})
return _mm256_mask_reduce_mul_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_mul_epi8((__mmask32)0b11111111111111110000000000000000,
(__m256i)(__v32qs){1,1,1,1,1,1,1,2,1,1,1,1,1,1,1,2,1,1,1,1,1,1,1,2,1,1,1,1,1,1,1,2})
== 4);
+TEST_CONSTEXPR(_mm256_mask_reduce_mul_epi8((__mmask32)0b00000000000000001111111111111111,
(__m256i)(__v32qs){4,1,1,1,1,1,1,2,1,1,1,1,1,1,1,2,1,1,1,1,1,1,1,2,1,1,1,1,1,1,1,2})
== 16);
signed char test_mm256_mask_reduce_and_epi8(__mmask32 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_and_epi8
@@ -209,6 +254,8 @@ signed char test_mm256_mask_reduce_and_epi8(__mmask32 __M,
__m256i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.and.v32i8(<32 x i8> %{{.*}})
return _mm256_mask_reduce_and_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_and_epi8((__mmask32)0b11111111111111110000000000000000,
(__m256i)(__v32qs){1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0})
== 0);
+TEST_CONSTEXPR(_mm256_mask_reduce_and_epi8((__mmask32)0b00000000000000001111111111111111,
(__m256i)(__v32qs){1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0})
== 1);
signed char test_mm256_mask_reduce_or_epi8(__mmask32 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_or_epi8
@@ -216,30 +263,36 @@ signed char test_mm256_mask_reduce_or_epi8(__mmask32 __M,
__m256i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.or.v32i8(<32 x i8> %{{.*}})
return _mm256_mask_reduce_or_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_or_epi8((__mmask32)0b11111111111111110000000000000000,
(__m256i)(__v32qs){1,2,4,8,16,32,64,127,1,2,4,8,16,32,64,127,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0})
== 0);
+TEST_CONSTEXPR(_mm256_mask_reduce_or_epi8((__mmask32)0b00000000000000001111111111111111,
(__m256i)(__v32qs){1,2,4,8,16,32,64,127,1,2,4,8,16,32,64,127,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0})
== 127);
short test_mm_reduce_max_epi16(__m128i __W){
// CHECK-LABEL: test_mm_reduce_max_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.smax.v8i16(<8 x i16> %{{.*}})
return _mm_reduce_max_epi16(__W);
}
+TEST_CONSTEXPR(_mm_reduce_max_epi16((__m128i)(__v8hi){-4,-3,-2,-1,1,2,3,4}) ==
4);
short test_mm_reduce_min_epi16(__m128i __W){
// CHECK-LABEL: test_mm_reduce_min_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.smin.v8i16(<8 x i16> %{{.*}})
return _mm_reduce_min_epi16(__W);
}
+TEST_CONSTEXPR(_mm_reduce_min_epi16((__m128i)(__v8hi){-4,-3,-2,-1,1,2,3,4}) ==
-4);
unsigned short test_mm_reduce_max_epu16(__m128i __W){
// CHECK-LABEL: test_mm_reduce_max_epu16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.umax.v8i16(<8 x i16> %{{.*}})
return _mm_reduce_max_epu16(__W);
}
+TEST_CONSTEXPR(_mm_reduce_max_epu16((__m128i)(__v8hu){1,2,3,4,5,6,7,8}) == 8);
unsigned short test_mm_reduce_min_epu16(__m128i __W){
// CHECK-LABEL: test_mm_reduce_min_epu16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.umin.v8i16(<8 x i16> %{{.*}})
return _mm_reduce_min_epu16(__W);
}
+TEST_CONSTEXPR(_mm_reduce_min_epu16((__m128i)(__v8hu){1,2,3,4,5,6,7,8}) == 1);
short test_mm_mask_reduce_max_epi16(__mmask8 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_max_epi16
@@ -247,6 +300,8 @@ short test_mm_mask_reduce_max_epi16(__mmask8 __M, __m128i
__W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.smax.v8i16(<8 x i16> %{{.*}})
return _mm_mask_reduce_max_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_max_epi16((__mmask8)0b11110000,
(__m128i)(__v8hi){-4,-3,-2,-1,1,2,3,4}) == 4);
+TEST_CONSTEXPR(_mm_mask_reduce_max_epi16((__mmask8)0b00001111,
(__m128i)(__v8hi){-4,-3,-2,-1,1,2,3,4}) == -1);
short test_mm_mask_reduce_min_epi16(__mmask8 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_min_epi16
@@ -254,6 +309,8 @@ short test_mm_mask_reduce_min_epi16(__mmask8 __M, __m128i
__W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.smin.v8i16(<8 x i16> %{{.*}})
return _mm_mask_reduce_min_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_min_epi16((__mmask8)0b11110000,
(__m128i)(__v8hi){-4,-3,-2,-1,1,2,3,4}) == 1);
+TEST_CONSTEXPR(_mm_mask_reduce_min_epi16((__mmask8)0b00001111,
(__m128i)(__v8hi){-4,-3,-2,-1,1,2,3,4}) == -4);
unsigned short test_mm_mask_reduce_max_epu16(__mmask8 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_max_epu16
@@ -261,6 +318,8 @@ unsigned short test_mm_mask_reduce_max_epu16(__mmask8 __M,
__m128i __W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.umax.v8i16(<8 x i16> %{{.*}})
return _mm_mask_reduce_max_epu16(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_max_epu16((__mmask8)0b11110000,
(__m128i)(__v8hu){1,2,3,4,5,6,7,8}) == 8);
+TEST_CONSTEXPR(_mm_mask_reduce_max_epu16((__mmask8)0b00001111,
(__m128i)(__v8hu){1,2,3,4,5,6,7,8}) == 4);
unsigned short test_mm_mask_reduce_min_epu16(__mmask8 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_min_epu16
@@ -268,30 +327,36 @@ unsigned short test_mm_mask_reduce_min_epu16(__mmask8
__M, __m128i __W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.umin.v8i16(<8 x i16> %{{.*}})
return _mm_mask_reduce_min_epu16(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_min_epu16((__mmask8)0b11110000,
(__m128i)(__v8hu){1,2,3,4,5,6,7,8}) == 5);
+TEST_CONSTEXPR(_mm_mask_reduce_min_epu16((__mmask8)0b00001111,
(__m128i)(__v8hu){1,2,3,4,5,6,7,8}) == 1);
short test_mm256_reduce_max_epi16(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_max_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.smax.v16i16(<16 x i16> %{{.*}})
return _mm256_reduce_max_epi16(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_max_epi16((__m256i)(__v16hi){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8})
== 8);
short test_mm256_reduce_min_epi16(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_min_epi16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.smin.v16i16(<16 x i16> %{{.*}})
return _mm256_reduce_min_epi16(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_min_epi16((__m256i)(__v16hi){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8})
== -8);
unsigned short test_mm256_reduce_max_epu16(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_max_epu16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.umax.v16i16(<16 x i16> %{{.*}})
return _mm256_reduce_max_epu16(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_max_epu16((__m256i)(__v16hu){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16})
== 16);
unsigned short test_mm256_reduce_min_epu16(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_min_epu16
// CHECK: call {{.*}}i16 @llvm.vector.reduce.umin.v16i16(<16 x i16> %{{.*}})
return _mm256_reduce_min_epu16(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_min_epu16((__m256i)(__v16hu){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16})
== 1);
short test_mm256_mask_reduce_max_epi16(__mmask16 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_max_epi16
@@ -299,6 +364,8 @@ short test_mm256_mask_reduce_max_epi16(__mmask16 __M,
__m256i __W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.smax.v16i16(<16 x i16> %{{.*}})
return _mm256_mask_reduce_max_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_max_epi16((__mmask16){0b1111111100000000},
(__m256i)(__v16hi){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8}) == 8);
+TEST_CONSTEXPR(_mm256_mask_reduce_max_epi16((__mmask16){0b0000000011111111},
(__m256i)(__v16hi){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8}) == -1);
short test_mm256_mask_reduce_min_epi16(__mmask16 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_min_epi16
@@ -306,6 +373,8 @@ short test_mm256_mask_reduce_min_epi16(__mmask16 __M,
__m256i __W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.smin.v16i16(<16 x i16> %{{.*}})
return _mm256_mask_reduce_min_epi16(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_min_epi16((__mmask16){0b1111111100000000},
(__m256i)(__v16hi){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8}) == 1);
+TEST_CONSTEXPR(_mm256_mask_reduce_min_epi16((__mmask16){0b0000000011111111},
(__m256i)(__v16hi){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8}) == -8);
unsigned short test_mm256_mask_reduce_max_epu16(__mmask16 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_max_epu16
@@ -313,6 +382,8 @@ unsigned short test_mm256_mask_reduce_max_epu16(__mmask16
__M, __m256i __W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.umax.v16i16(<16 x i16> %{{.*}})
return _mm256_mask_reduce_max_epu16(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_max_epu16((__mmask16){0b1111111100000000},
(__m256i)(__v16hu){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}) == 16);
+TEST_CONSTEXPR(_mm256_mask_reduce_max_epu16((__mmask16){0b0000000011111111},
(__m256i)(__v16hu){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}) == 8);
unsigned short test_mm256_mask_reduce_min_epu16(__mmask16 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_min_epu16
@@ -320,30 +391,36 @@ unsigned short test_mm256_mask_reduce_min_epu16(__mmask16
__M, __m256i __W){
// CHECK: call {{.*}}i16 @llvm.vector.reduce.umin.v16i16(<16 x i16> %{{.*}})
return _mm256_mask_reduce_min_epu16(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_min_epu16((__mmask16){0b1111111100000000},
(__m256i)(__v16hu){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}) == 9);
+TEST_CONSTEXPR(_mm256_mask_reduce_min_epu16((__mmask16){0b0000000011111111},
(__m256i)(__v16hu){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}) == 1);
signed char test_mm_reduce_max_epi8(__m128i __W){
// CHECK-LABEL: test_mm_reduce_max_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> %{{.*}})
return _mm_reduce_max_epi8(__W);
}
+TEST_CONSTEXPR(_mm_reduce_max_epi8((__m128i)(__v16qs){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8})
== 8);
signed char test_mm_reduce_min_epi8(__m128i __W){
// CHECK-LABEL: test_mm_reduce_min_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.smin.v16i8(<16 x i8> %{{.*}})
return _mm_reduce_min_epi8(__W);
}
+TEST_CONSTEXPR(_mm_reduce_min_epi8((__m128i)(__v16qs){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8})
== -8);
unsigned char test_mm_reduce_max_epu8(__m128i __W){
// CHECK-LABEL: test_mm_reduce_max_epu8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.umax.v16i8(<16 x i8> %{{.*}})
return _mm_reduce_max_epu8(__W);
}
+TEST_CONSTEXPR(_mm_reduce_max_epu8((__m128i)(__v16qu){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16})
== 16);
unsigned char test_mm_reduce_min_epu8(__m128i __W){
// CHECK-LABEL: test_mm_reduce_min_epu8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.umin.v16i8(<16 x i8> %{{.*}})
return _mm_reduce_min_epu8(__W);
}
+TEST_CONSTEXPR(_mm_reduce_min_epu8((__m128i)(__v16qu){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16})
== 1);
signed char test_mm_mask_reduce_max_epi8(__mmask16 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_max_epi8
@@ -351,6 +428,8 @@ signed char test_mm_mask_reduce_max_epi8(__mmask16 __M,
__m128i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> %{{.*}})
return _mm_mask_reduce_max_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_max_epi8((__mmask16)0b1111111100000000,
(__m128i)(__v16qs){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8}) == 8);
+TEST_CONSTEXPR(_mm_mask_reduce_max_epi8((__mmask16)0b0000000011111111,
(__m128i)(__v16qs){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8}) == -1);
signed char test_mm_mask_reduce_min_epi8(__mmask16 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_min_epi8
@@ -358,6 +437,8 @@ signed char test_mm_mask_reduce_min_epi8(__mmask16 __M,
__m128i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.smin.v16i8(<16 x i8> %{{.*}})
return _mm_mask_reduce_min_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_min_epi8((__mmask16)0b1111111100000000,
(__m128i)(__v16qs){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8}) == 1);
+TEST_CONSTEXPR(_mm_mask_reduce_min_epi8((__mmask16)0b0000000011111111,
(__m128i)(__v16qs){-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8}) == -8);
unsigned char test_mm_mask_reduce_max_epu8(__mmask16 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_max_epu8
@@ -365,6 +446,8 @@ unsigned char test_mm_mask_reduce_max_epu8(__mmask16 __M,
__m128i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.umax.v16i8(<16 x i8> %{{.*}})
return _mm_mask_reduce_max_epu8(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_max_epu8((__mmask16)0b1111111100000000,
(__m128i)(__v16qs){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}) == 16);
+TEST_CONSTEXPR(_mm_mask_reduce_max_epu8((__mmask16)0b0000000011111111,
(__m128i)(__v16qs){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}) == 8);
unsigned char test_mm_mask_reduce_min_epu8(__mmask16 __M, __m128i __W){
// CHECK-LABEL: test_mm_mask_reduce_min_epu8
@@ -372,30 +455,36 @@ unsigned char test_mm_mask_reduce_min_epu8(__mmask16 __M,
__m128i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.umin.v16i8(<16 x i8> %{{.*}})
return _mm_mask_reduce_min_epu8(__M, __W);
}
+TEST_CONSTEXPR(_mm_mask_reduce_min_epu8((__mmask16)0b1111111100000000,
(__m128i)(__v16qs){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}) == 9);
+TEST_CONSTEXPR(_mm_mask_reduce_min_epu8((__mmask16)0b0000000011111111,
(__m128i)(__v16qs){1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}) == 1);
signed char test_mm256_reduce_max_epi8(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_max_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.smax.v32i8(<32 x i8> %{{.*}})
return _mm256_reduce_max_epi8(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_max_epi8((__m256i)(__v32qs){-16,-15,-14,-13,-12,-11,-10,-9,-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16})
== 16);
signed char test_mm256_reduce_min_epi8(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_min_epi8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.smin.v32i8(<32 x i8> %{{.*}})
return _mm256_reduce_min_epi8(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_min_epi8((__m256i)(__v32qs){-16,-15,-14,-13,-12,-11,-10,-9,-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16})
== -16);
unsigned char test_mm256_reduce_max_epu8(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_max_epu8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.umax.v32i8(<32 x i8> %{{.*}})
return _mm256_reduce_max_epu8(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_max_epu8((__m256i)(__v32qu){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})
== 32);
unsigned char test_mm256_reduce_min_epu8(__m256i __W){
// CHECK-LABEL: test_mm256_reduce_min_epu8
// CHECK: call {{.*}}i8 @llvm.vector.reduce.umin.v32i8(<32 x i8> %{{.*}})
return _mm256_reduce_min_epu8(__W);
}
+TEST_CONSTEXPR(_mm256_reduce_min_epu8((__m256i)(__v32qu){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})
== 1);
signed char test_mm256_mask_reduce_max_epi8(__mmask32 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_max_epi8
@@ -403,6 +492,8 @@ signed char test_mm256_mask_reduce_max_epi8(__mmask32 __M,
__m256i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.smax.v32i8(<32 x i8> %{{.*}})
return _mm256_mask_reduce_max_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_max_epi8((__mmask32)0b11111111111111110000000000000000,
(__m256i)(__v32qs){-16,-15,-14,-13,-12,-11,-10,-9,-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16})
== 16);
+TEST_CONSTEXPR(_mm256_mask_reduce_max_epi8((__mmask32)0b00000000000000001111111111111111,
(__m256i)(__v32qs){-16,-15,-14,-13,-12,-11,-10,-9,-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16})
== -1);
signed char test_mm256_mask_reduce_min_epi8(__mmask32 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_min_epi8
@@ -410,6 +501,8 @@ signed char test_mm256_mask_reduce_min_epi8(__mmask32 __M,
__m256i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.smin.v32i8(<32 x i8> %{{.*}})
return _mm256_mask_reduce_min_epi8(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_min_epi8((__mmask32)0b11111111111111110000000000000000,
(__m256i)(__v32qs){-16,-15,-14,-13,-12,-11,-10,-9,-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16})
== 1);
+TEST_CONSTEXPR(_mm256_mask_reduce_min_epi8((__mmask32)0b00000000000000001111111111111111,
(__m256i)(__v32qs){-16,-15,-14,-13,-12,-11,-10,-9,-8,-7,-6,-5,-4,-3,-2,-1,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16})
== -16);
unsigned char test_mm256_mask_reduce_max_epu8(__mmask32 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_max_epu8
@@ -417,6 +510,8 @@ unsigned char test_mm256_mask_reduce_max_epu8(__mmask32
__M, __m256i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.umax.v32i8(<32 x i8> %{{.*}})
return _mm256_mask_reduce_max_epu8(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_max_epu8((__mmask32)0b11111111111111110000000000000000,
(__m256i)(__v32qu){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})
== 32);
+TEST_CONSTEXPR(_mm256_mask_reduce_max_epu8((__mmask32)0b00000000000000001111111111111111,
(__m256i)(__v32qu){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})
== 16);
unsigned char test_mm256_mask_reduce_min_epu8(__mmask32 __M, __m256i __W){
// CHECK-LABEL: test_mm256_mask_reduce_min_epu8
@@ -424,3 +519,5 @@ unsigned char test_mm256_mask_reduce_min_epu8(__mmask32
__M, __m256i __W){
// CHECK: call {{.*}}i8 @llvm.vector.reduce.umin.v32i8(<32 x i8> %{{.*}})
return _mm256_mask_reduce_min_epu8(__M, __W);
}
+TEST_CONSTEXPR(_mm256_mask_reduce_min_epu8((__mmask32)0b11111111111111110000000000000000,
(__m256i)(__v32qu){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})
== 17);
+TEST_CONSTEXPR(_mm256_mask_reduce_min_epu8((__mmask32)0b00000000000000001111111111111111,
(__m256i)(__v32qu){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})
== 1);
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
