stk-code_catmod/lib/simd_wrapper/simde/x86/sse3.h
2022-04-29 11:02:25 +08:00

500 lines
16 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:
* 2017-2020 Evan Nemerson <evan@nemerson.com>
*/
#if !defined(SIMDE_X86_SSE3_H)
#define SIMDE_X86_SSE3_H
#include "sse2.h"
HEDLEY_DIAGNOSTIC_PUSH
SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
SIMDE_BEGIN_DECLS_
SIMDE_FUNCTION_ATTRIBUTES
simde__m128i
simde_x_mm_deinterleaveeven_epi16 (simde__m128i a, simde__m128i b) {
simde__m128i_private
r_,
a_ = simde__m128i_to_private(a),
b_ = simde__m128i_to_private(b);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_i16 = vuzp1q_s16(a_.neon_i16, b_.neon_i16);
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
int16x8x2_t t = vuzpq_s16(a_.neon_i16, b_.neon_i16);
r_.neon_i16 = t.val[0];
#elif defined(SIMDE_SHUFFLE_VECTOR_)
r_.i16 = SIMDE_SHUFFLE_VECTOR_(16, 16, a_.i16, b_.i16, 0, 2, 4, 6, 8, 10, 12, 14);
#else
const size_t halfway_point = (sizeof(r_.i16) / sizeof(r_.i16[0])) / 2;
for(size_t i = 0 ; i < halfway_point ; i++) {
r_.i16[i] = a_.i16[2 * i];
r_.i16[i + halfway_point] = b_.i16[2 * i];
}
#endif
return simde__m128i_from_private(r_);
}
SIMDE_FUNCTION_ATTRIBUTES
simde__m128i
simde_x_mm_deinterleaveodd_epi16 (simde__m128i a, simde__m128i b) {
simde__m128i_private
r_,
a_ = simde__m128i_to_private(a),
b_ = simde__m128i_to_private(b);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_i16 = vuzp2q_s16(a_.neon_i16, b_.neon_i16);
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
int16x8x2_t t = vuzpq_s16(a_.neon_i16, b_.neon_i16);
r_.neon_i16 = t.val[1];
#elif defined(SIMDE_SHUFFLE_VECTOR_)
r_.i16 = SIMDE_SHUFFLE_VECTOR_(16, 16, a_.i16, b_.i16, 1, 3, 5, 7, 9, 11, 13, 15);
#else
const size_t halfway_point = (sizeof(r_.i16) / sizeof(r_.i16[0])) / 2;
for(size_t i = 0 ; i < halfway_point ; i++) {
r_.i16[i] = a_.i16[2 * i + 1];
r_.i16[i + halfway_point] = b_.i16[2 * i + 1];
}
#endif
return simde__m128i_from_private(r_);
}
SIMDE_FUNCTION_ATTRIBUTES
simde__m128i
simde_x_mm_deinterleaveeven_epi32 (simde__m128i a, simde__m128i b) {
simde__m128i_private
r_,
a_ = simde__m128i_to_private(a),
b_ = simde__m128i_to_private(b);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_i32 = vuzp1q_s32(a_.neon_i32, b_.neon_i32);
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
int32x4x2_t t = vuzpq_s32(a_.neon_i32, b_.neon_i32);
r_.neon_i32 = t.val[0];
#elif defined(SIMDE_SHUFFLE_VECTOR_)
r_.i32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.i32, b_.i32, 0, 2, 4, 6);
#else
const size_t halfway_point = (sizeof(r_.i32) / sizeof(r_.i32[0])) / 2;
for(size_t i = 0 ; i < halfway_point ; i++) {
r_.i32[i] = a_.i32[2 * i];
r_.i32[i + halfway_point] = b_.i32[2 * i];
}
#endif
return simde__m128i_from_private(r_);
}
SIMDE_FUNCTION_ATTRIBUTES
simde__m128i
simde_x_mm_deinterleaveodd_epi32 (simde__m128i a, simde__m128i b) {
simde__m128i_private
r_,
a_ = simde__m128i_to_private(a),
b_ = simde__m128i_to_private(b);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_i32 = vuzp2q_s32(a_.neon_i32, b_.neon_i32);
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
int32x4x2_t t = vuzpq_s32(a_.neon_i32, b_.neon_i32);
r_.neon_i32 = t.val[1];
#elif defined(SIMDE_SHUFFLE_VECTOR_)
r_.i32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.i32, b_.i32, 1, 3, 5, 7);
#else
const size_t halfway_point = (sizeof(r_.i32) / sizeof(r_.i32[0])) / 2;
for(size_t i = 0 ; i < halfway_point ; i++) {
r_.i32[i] = a_.i32[2 * i + 1];
r_.i32[i + halfway_point] = b_.i32[2 * i + 1];
}
#endif
return simde__m128i_from_private(r_);
}
SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_x_mm_deinterleaveeven_ps (simde__m128 a, simde__m128 b) {
simde__m128_private
r_,
a_ = simde__m128_to_private(a),
b_ = simde__m128_to_private(b);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_f32 = vuzp1q_f32(a_.neon_f32, b_.neon_f32);
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
float32x4x2_t t = vuzpq_f32(a_.neon_f32, b_.neon_f32);
r_.neon_f32 = t.val[0];
#elif defined(SIMDE_SHUFFLE_VECTOR_)
r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 0, 2, 4, 6);
#else
const size_t halfway_point = (sizeof(r_.f32) / sizeof(r_.f32[0])) / 2;
for(size_t i = 0 ; i < halfway_point ; i++) {
r_.f32[i] = a_.f32[2 * i];
r_.f32[i + halfway_point] = b_.f32[2 * i];
}
#endif
return simde__m128_from_private(r_);
}
SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_x_mm_deinterleaveodd_ps (simde__m128 a, simde__m128 b) {
simde__m128_private
r_,
a_ = simde__m128_to_private(a),
b_ = simde__m128_to_private(b);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_f32 = vuzp2q_f32(a_.neon_f32, b_.neon_f32);
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
float32x4x2_t t = vuzpq_f32(a_.neon_f32, b_.neon_f32);
r_.neon_f32 = t.val[1];
#elif defined(SIMDE_SHUFFLE_VECTOR_)
r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 1, 3, 5, 7);
#else
const size_t halfway_point = (sizeof(r_.f32) / sizeof(r_.f32[0])) / 2;
for(size_t i = 0 ; i < halfway_point ; i++) {
r_.f32[i] = a_.f32[2 * i + 1];
r_.f32[i + halfway_point] = b_.f32[2 * i + 1];
}
#endif
return simde__m128_from_private(r_);
}
SIMDE_FUNCTION_ATTRIBUTES
simde__m128d
simde_x_mm_deinterleaveeven_pd (simde__m128d a, simde__m128d b) {
simde__m128d_private
r_,
a_ = simde__m128d_to_private(a),
b_ = simde__m128d_to_private(b);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_f64 = vuzp1q_f64(a_.neon_f64, b_.neon_f64);
#elif defined(SIMDE_SHUFFLE_VECTOR_)
r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, b_.f64, 0, 2);
#else
const size_t halfway_point = (sizeof(r_.f64) / sizeof(r_.f64[0])) / 2;
for(size_t i = 0 ; i < halfway_point ; i++) {
r_.f64[i] = a_.f64[2 * i];
r_.f64[i + halfway_point] = b_.f64[2 * i];
}
#endif
return simde__m128d_from_private(r_);
}
SIMDE_FUNCTION_ATTRIBUTES
simde__m128d
simde_x_mm_deinterleaveodd_pd (simde__m128d a, simde__m128d b) {
simde__m128d_private
r_,
a_ = simde__m128d_to_private(a),
b_ = simde__m128d_to_private(b);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_f64 = vuzp2q_f64(a_.neon_f64, b_.neon_f64);
#elif defined(SIMDE_SHUFFLE_VECTOR_)
r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, b_.f64, 1, 3);
#else
const size_t halfway_point = (sizeof(r_.f64) / sizeof(r_.f64[0])) / 2;
for(size_t i = 0 ; i < halfway_point ; i++) {
r_.f64[i] = a_.f64[2 * i + 1];
r_.f64[i + halfway_point] = b_.f64[2 * i + 1];
}
#endif
return simde__m128d_from_private(r_);
}
SIMDE_FUNCTION_ATTRIBUTES
simde__m128d
simde_mm_addsub_pd (simde__m128d a, simde__m128d b) {
#if defined(SIMDE_X86_SSE3_NATIVE)
return _mm_addsub_pd(a, b);
#else
simde__m128d_private
r_,
a_ = simde__m128d_to_private(a),
b_ = simde__m128d_to_private(b);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
float64x2_t rs = vsubq_f64(a_.neon_f64, b_.neon_f64);
float64x2_t ra = vaddq_f64(a_.neon_f64, b_.neon_f64);
return vcombine_f64(vget_low_f64(rs), vget_high_f64(ra));
#elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_SHUFFLE_VECTOR_)
r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64 - b_.f64, a_.f64 + b_.f64, 0, 3);
#else
for (size_t i = 0 ; i < (sizeof(r_.f64) / sizeof(r_.f64[0])) ; i += 2) {
r_.f64[ i ] = a_.f64[ i ] - b_.f64[ i ];
r_.f64[1 + i] = a_.f64[1 + i] + b_.f64[1 + i];
}
#endif
return simde__m128d_from_private(r_);
#endif
}
#if defined(SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES)
# define _mm_addsub_pd(a, b) simde_mm_addsub_pd(a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_addsub_ps (simde__m128 a, simde__m128 b) {
#if defined(SIMDE_X86_SSE3_NATIVE)
return _mm_addsub_ps(a, b);
#else
simde__m128_private
r_,
a_ = simde__m128_to_private(a),
b_ = simde__m128_to_private(b);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
float32x4_t rs = vsubq_f32(a_.neon_f32, b_.neon_f32);
float32x4_t ra = vaddq_f32(a_.neon_f32, b_.neon_f32);
return vtrn2q_f32(vreinterpretq_f32_s32(vrev64q_s32(vreinterpretq_s32_f32(rs))), ra);
#elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_SHUFFLE_VECTOR_)
r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32 - b_.f32, a_.f32 + b_.f32, 0, 5, 2, 7);
#else
for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i += 2) {
r_.f32[ i ] = a_.f32[ i ] - b_.f32[ i ];
r_.f32[1 + i] = a_.f32[1 + i] + b_.f32[1 + i];
}
#endif
return simde__m128_from_private(r_);
#endif
}
#if defined(SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES)
# define _mm_addsub_ps(a, b) simde_mm_addsub_ps(a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m128d
simde_mm_hadd_pd (simde__m128d a, simde__m128d b) {
#if defined(SIMDE_X86_SSE3_NATIVE)
return _mm_hadd_pd(a, b);
#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
return simde__m128d_from_neon_f64(vpaddq_f64(simde__m128d_to_neon_f64(a), simde__m128d_to_neon_f64(b)));
#else
return simde_mm_add_pd(simde_x_mm_deinterleaveeven_pd(a, b), simde_x_mm_deinterleaveodd_pd(a, b));
#endif
}
#if defined(SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES)
# define _mm_hadd_pd(a, b) simde_mm_hadd_pd(a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_hadd_ps (simde__m128 a, simde__m128 b) {
#if defined(SIMDE_X86_SSE3_NATIVE)
return _mm_hadd_ps(a, b);
#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
return simde__m128_from_neon_f32(vpaddq_f32(simde__m128_to_neon_f32(a), simde__m128_to_neon_f32(b)));
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
float32x4x2_t t = vuzpq_f32(simde__m128_to_neon_f32(a), simde__m128_to_neon_f32(b));
return simde__m128_from_neon_f32(vaddq_f32(t.val[0], t.val[1]));
#else
return simde_mm_add_ps(simde_x_mm_deinterleaveeven_ps(a, b), simde_x_mm_deinterleaveodd_ps(a, b));
#endif
}
#if defined(SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES)
# define _mm_hadd_ps(a, b) simde_mm_hadd_ps(a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m128d
simde_mm_hsub_pd (simde__m128d a, simde__m128d b) {
#if defined(SIMDE_X86_SSE3_NATIVE)
return _mm_hsub_pd(a, b);
#else
return simde_mm_sub_pd(simde_x_mm_deinterleaveeven_pd(a, b), simde_x_mm_deinterleaveodd_pd(a, b));
#endif
}
#if defined(SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES)
# define _mm_hsub_pd(a, b) simde_mm_hsub_pd(a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_hsub_ps (simde__m128 a, simde__m128 b) {
#if defined(SIMDE_X86_SSE3_NATIVE)
return _mm_hsub_ps(a, b);
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
float32x4x2_t t = vuzpq_f32(simde__m128_to_neon_f32(a), simde__m128_to_neon_f32(b));
return simde__m128_from_neon_f32(vaddq_f32(t.val[0], vnegq_f32(t.val[1])));
#else
return simde_mm_sub_ps(simde_x_mm_deinterleaveeven_ps(a, b), simde_x_mm_deinterleaveodd_ps(a, b));
#endif
}
#if defined(SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES)
# define _mm_hsub_ps(a, b) simde_mm_hsub_ps(a, b)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m128i
simde_mm_lddqu_si128 (simde__m128i const* mem_addr) {
#if defined(SIMDE_X86_SSE3_NATIVE)
return _mm_lddqu_si128(mem_addr);
#else
simde__m128i_private r_;
#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
r_.neon_i32 = vld1q_s32(HEDLEY_REINTERPRET_CAST(int32_t const*, mem_addr));
#else
simde_memcpy(&r_, mem_addr, sizeof(r_));
#endif
return simde__m128i_from_private(r_);
#endif
}
#if defined(SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES)
# define _mm_lddqu_si128(mem_addr) simde_mm_lddqu_si128(mem_addr)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m128d
simde_mm_loaddup_pd (simde_float64 const* mem_addr) {
#if defined(SIMDE_X86_SSE3_NATIVE)
return _mm_loaddup_pd(mem_addr);
#else
simde__m128d_private r_;
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_f64 = vdupq_n_f64(*mem_addr);
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
r_.neon_i64 = vdupq_n_s64(*HEDLEY_REINTERPRET_CAST(int64_t const*, mem_addr));
#else
r_.f64[0] = *mem_addr;
r_.f64[1] = *mem_addr;
#endif
return simde__m128d_from_private(r_);
#endif
}
#if defined(SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES)
# define _mm_loaddup_pd(mem_addr) simde_mm_loaddup_pd(mem_addr)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m128d
simde_mm_movedup_pd (simde__m128d a) {
#if defined(SIMDE_X86_SSE3_NATIVE)
return _mm_movedup_pd(a);
#else
simde__m128d_private
r_,
a_ = simde__m128d_to_private(a);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_f64 = vdupq_laneq_f64(a_.neon_f64, 0);
#elif defined(SIMDE_WASM_SIMD128_NATIVE)
r_.wasm_v128 = wasm_i64x2_shuffle(a_.wasm_v128, a_.wasm_v128, 0, 0);
#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && defined(SIMDE_SHUFFLE_VECTOR_)
r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, a_.f64, 0, 0);
#else
r_.f64[0] = a_.f64[0];
r_.f64[1] = a_.f64[0];
#endif
return simde__m128d_from_private(r_);
#endif
}
#if defined(SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES)
# define _mm_movedup_pd(a) simde_mm_movedup_pd(a)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_movehdup_ps (simde__m128 a) {
#if defined(SIMDE_X86_SSE3_NATIVE)
return _mm_movehdup_ps(a);
#else
simde__m128_private
r_,
a_ = simde__m128_to_private(a);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_f32 = vtrn2q_f32(a_.neon_f32, a_.neon_f32);
#elif defined(SIMDE_WASM_SIMD128_NATIVE)
r_.wasm_v128 = wasm_i32x4_shuffle(a_.wasm_v128, a_.wasm_v128, 1, 1, 3, 3);
#elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_SHUFFLE_VECTOR_)
r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, a_.f32, 1, 1, 3, 3);
#else
r_.f32[0] = a_.f32[1];
r_.f32[1] = a_.f32[1];
r_.f32[2] = a_.f32[3];
r_.f32[3] = a_.f32[3];
#endif
return simde__m128_from_private(r_);
#endif
}
#if defined(SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES)
# define _mm_movehdup_ps(a) simde_mm_movehdup_ps(a)
#endif
SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_moveldup_ps (simde__m128 a) {
#if defined(SIMDE__SSE3_NATIVE)
return _mm_moveldup_ps(a);
#else
simde__m128_private
r_,
a_ = simde__m128_to_private(a);
#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
r_.neon_f32 = vtrn1q_f32(a_.neon_f32, a_.neon_f32);
#elif defined(SIMDE_WASM_SIMD128_NATIVE)
r_.wasm_v128 = wasm_i32x4_shuffle(a_.wasm_v128, a_.wasm_v128, 0, 0, 2, 2);
#elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_SHUFFLE_VECTOR_)
r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, a_.f32, 0, 0, 2, 2);
#else
r_.f32[0] = a_.f32[0];
r_.f32[1] = a_.f32[0];
r_.f32[2] = a_.f32[2];
r_.f32[3] = a_.f32[2];
#endif
return simde__m128_from_private(r_);
#endif
}
#if defined(SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES)
# define _mm_moveldup_ps(a) simde_mm_moveldup_ps(a)
#endif
SIMDE_END_DECLS_
HEDLEY_DIAGNOSTIC_POP
#endif /* !defined(SIMDE_X86_SSE3_H) */