stk-code_catmod/lib/simd_wrapper/simde/x86/avx512/mul.h
2022-04-29 11:02:25 +08:00

280 lines
8.9 KiB
C

/* SPDX-License-Identifier: MIT
*
* Permission is hereby granted, free of charge, to any person
* obtaining a copy of this software and associated documentation
* files (the "Software"), to deal in the Software without
* restriction, including without limitation the rights to use, copy,
* modify, merge, publish, distribute, sublicense, and/or sell copies
* of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be
* included in all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*
* Copyright:
* 2020 Evan Nemerson <evan@nemerson.com>
* 2020 Himanshi Mathur <himanshi18037@iiitd.ac.in>
* 2020 Hidayat Khan <huk2209@gmail.com>
*/
#if !defined(SIMDE_X86_AVX512_MUL_H)
#define SIMDE_X86_AVX512_MUL_H
#include "types.h"
#include "mov.h"
HEDLEY_DIAGNOSTIC_PUSH
SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
SIMDE_BEGIN_DECLS_
SIMDE_FUNCTION_ATTRIBUTES
simde__m512
simde_mm512_mul_ps (simde__m512 a, simde__m512 b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_mul_ps(a, b);
#else
simde__m512_private
r_,
a_ = simde__m512_to_private(a),
b_ = simde__m512_to_private(b);
#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
r_.f32 = a_.f32 * b_.f32;
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.m256) / sizeof(r_.m256[0])) ; i++) {
r_.m256[i] = simde_mm256_mul_ps(a_.m256[i], b_.m256[i]);
}
#endif
return simde__m512_from_private(r_);
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_mul_ps
#define _mm512_mul_ps(a, b) simde_mm512_mul_ps(a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m512
simde_mm512_mask_mul_ps(simde__m512 src, simde__mmask16 k, simde__m512 a, simde__m512 b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_mask_mul_ps(src, k, a, b);
#else
return simde_mm512_mask_mov_ps(src, k, simde_mm512_mul_ps(a, b));
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_mask_mul_ps
#define _mm512_mask_mul_ps(src, k, a, b) simde_mm512_mask_mul_ps(src, k, a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m512
simde_mm512_maskz_mul_ps(simde__mmask16 k, simde__m512 a, simde__m512 b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_maskz_mul_ps(k, a, b);
#else
return simde_mm512_maskz_mov_ps(k, simde_mm512_mul_ps(a, b));
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_maskz_mul_ps
#define _mm512_maskz_mul_ps(k, a, b) simde_mm512_maskz_mul_ps(k, a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m512d
simde_mm512_mul_pd (simde__m512d a, simde__m512d b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_mul_pd(a, b);
#else
simde__m512d_private
r_,
a_ = simde__m512d_to_private(a),
b_ = simde__m512d_to_private(b);
#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
r_.f64 = a_.f64 * b_.f64;
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.m256d) / sizeof(r_.m256d[0])) ; i++) {
r_.m256d[i] = simde_mm256_mul_pd(a_.m256d[i], b_.m256d[i]);
}
#endif
return simde__m512d_from_private(r_);
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_mul_pd
#define _mm512_mul_pd(a, b) simde_mm512_mul_pd(a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m512d
simde_mm512_mask_mul_pd(simde__m512d src, simde__mmask8 k, simde__m512d a, simde__m512d b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_mask_mul_pd(src, k, a, b);
#else
return simde_mm512_mask_mov_pd(src, k, simde_mm512_mul_pd(a, b));
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_mask_mul_pd
#define _mm512_mask_mul_pd(src, k, a, b) simde_mm512_mask_mul_pd(src, k, a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m512d
simde_mm512_maskz_mul_pd(simde__mmask8 k, simde__m512d a, simde__m512d b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_maskz_mul_pd(k, a, b);
#else
return simde_mm512_maskz_mov_pd(k, simde_mm512_mul_pd(a, b));
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_maskz_mul_pd
#define _mm512_maskz_mul_pd(k, a, b) simde_mm512_maskz_mul_pd(k, a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m512i
simde_mm512_mul_epi32 (simde__m512i a, simde__m512i b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_mul_epi32(a, b);
#else
simde__m512i_private
r_,
a_ = simde__m512i_to_private(a),
b_ = simde__m512i_to_private(b);
#if defined(SIMDE_CONVERT_VECTOR_) && defined(SIMDE_SHUFFLE_VECTOR_)
simde__m512i_private x;
__typeof__(r_.i64) ta, tb;
/* Get even numbered 32-bit values */
x.i32 = SIMDE_SHUFFLE_VECTOR_(32, 64, a_.i32, b_.i32, 0, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30);
/* Cast to 64 bits */
SIMDE_CONVERT_VECTOR_(ta, x.m256i_private[0].i32);
SIMDE_CONVERT_VECTOR_(tb, x.m256i_private[1].i32);
r_.i64 = ta * tb;
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.i64) / sizeof(r_.i64[0])) ; i++) {
r_.i64[i] = HEDLEY_STATIC_CAST(int64_t, a_.i32[i << 1]) * HEDLEY_STATIC_CAST(int64_t, b_.i32[i << 1]);
}
#endif
return simde__m512i_from_private(r_);
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_mul_epi32
#define _mm512_mul_epi32(a, b) simde_mm512_mul_epi32(a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m512i
simde_mm512_mask_mul_epi32(simde__m512i src, simde__mmask8 k, simde__m512i a, simde__m512i b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_mask_mul_epi32(src, k, a, b);
#else
return simde_mm512_mask_mov_epi64(src, k, simde_mm512_mul_epi32(a, b));
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_mask_mul_epi32
#define _mm512_mask_mul_epi32(src, k, a, b) simde_mm512_mask_mul_epi32(src, k, a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m512i
simde_mm512_maskz_mul_epi32(simde__mmask8 k, simde__m512i a, simde__m512i b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_maskz_mul_epi32(k, a, b);
#else
return simde_mm512_maskz_mov_epi64(k, simde_mm512_mul_epi32(a, b));
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_maskz_mul_epi32
#define _mm512_maskz_mul_epi32(k, a, b) simde_mm512_maskz_mul_epi32(k, a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m512i
simde_mm512_mul_epu32 (simde__m512i a, simde__m512i b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_mul_epu32(a, b);
#else
simde__m512i_private
r_,
a_ = simde__m512i_to_private(a),
b_ = simde__m512i_to_private(b);
#if defined(SIMDE_CONVERT_VECTOR_) && defined(SIMDE_SHUFFLE_VECTOR_)
simde__m512i_private x;
__typeof__(r_.u64) ta, tb;
x.u32 = SIMDE_SHUFFLE_VECTOR_(32, 64, a_.u32, b_.u32, 0, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30);
SIMDE_CONVERT_VECTOR_(ta, x.m256i_private[0].u32);
SIMDE_CONVERT_VECTOR_(tb, x.m256i_private[1].u32);
r_.u64 = ta * tb;
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.u64) / sizeof(r_.u64[0])) ; i++) {
r_.u64[i] = HEDLEY_STATIC_CAST(uint64_t, a_.u32[i << 1]) * HEDLEY_STATIC_CAST(uint64_t, b_.u32[i << 1]);
}
#endif
return simde__m512i_from_private(r_);
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_mul_epu32
#define _mm512_mul_epu32(a, b) simde_mm512_mul_epu32(a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m512i
simde_mm512_mask_mul_epu32(simde__m512i src, simde__mmask8 k, simde__m512i a, simde__m512i b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_mask_mul_epu32(src, k, a, b);
#else
return simde_mm512_mask_mov_epi64(src, k, simde_mm512_mul_epu32(a, b));
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_mask_mul_epu32
#define _mm512_mask_mul_epu32(src, k, a, b) simde_mm512_mask_mul_epu32(src, k, a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m512i
simde_mm512_maskz_mul_epu32(simde__mmask8 k, simde__m512i a, simde__m512i b) {
#if defined(SIMDE_X86_AVX512F_NATIVE)
return _mm512_maskz_mul_epu32(k, a, b);
#else
return simde_mm512_maskz_mov_epi64(k, simde_mm512_mul_epu32(a, b));
#endif
}
#if defined(SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES)
#undef _mm512_maskz_mul_epu32
#define _mm512_maskz_mul_epu32(k, a, b) simde_mm512_maskz_mul_epu32(k, a, b)
#endif
SIMDE_END_DECLS_
HEDLEY_DIAGNOSTIC_POP
#endif /* !defined(SIMDE_X86_AVX512_MUL_H) */