/* 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 * 2020 Christopher Moore */ /* The GFNI implementation is based on Wojciech Muła's work at * http://0x80.pl/articles/avx512-galois-field-for-bit-shuffling.html#bit-shuffling via * https://github.com/InstLatx64/InstLatX64_Demo/blob/49c27effdfd5a45f27e0ccb6e2f3be5f27c3845d/GFNI_Demo.h#L173 */ #if !defined(SIMDE_ARM_NEON_RBIT_H) #define SIMDE_ARM_NEON_RBIT_H #include "reinterpret.h" #include "types.h" HEDLEY_DIAGNOSTIC_PUSH SIMDE_DISABLE_UNWANTED_DIAGNOSTICS SIMDE_BEGIN_DECLS_ SIMDE_FUNCTION_ATTRIBUTES simde_uint8x8_t simde_vrbit_u8(simde_uint8x8_t a) { #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) return vrbit_u8(a); #elif defined(SIMDE_X86_MMX_NATIVE) && defined(SIMDE_X86_GFNI_NATIVE) __m128i tmp = _mm_movpi64_epi64(a); tmp = _mm_gf2p8affine_epi64_epi8(tmp, _mm_set1_epi64x(HEDLEY_STATIC_CAST(int64_t, UINT64_C(0x8040201008040201))), 0); return _mm_movepi64_pi64(tmp); #elif defined(SIMDE_X86_MMX_NATIVE) __m64 mask; mask = _mm_set1_pi8(0x55); a = _mm_or_si64(_mm_andnot_si64(mask, _mm_slli_pi16(a, 1)), _mm_and_si64(mask, _mm_srli_pi16(a, 1))); mask = _mm_set1_pi8(0x33); a = _mm_or_si64(_mm_andnot_si64(mask, _mm_slli_pi16(a, 2)), _mm_and_si64(mask, _mm_srli_pi16(a, 2))); mask = _mm_set1_pi8(0x0F); a = _mm_or_si64(_mm_andnot_si64(mask, _mm_slli_pi16(a, 4)), _mm_and_si64(mask, _mm_srli_pi16(a, 4))); return a; #else simde_uint8x8_private r_, a_ = simde_uint8x8_to_private(a); SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { #if HEDLEY_HAS_BUILTIN(__builtin_bitreverse8) && !defined(HEDLEY_IBM_VERSION) r_.values[i] = __builtin_bitreverse8(a_.values[i]); #else r_.values[i] = HEDLEY_STATIC_CAST(uint8_t, (((a_.values[i] * UINT64_C(0x80200802)) & UINT64_C(0x0884422110)) * UINT64_C(0x0101010101)) >> 32); #endif } return simde_uint8x8_from_private(r_); #endif } #if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) #undef vrbit_u8 #define vrbit_u8(a) simde_vrbit_u8(a) #endif SIMDE_FUNCTION_ATTRIBUTES simde_int8x8_t simde_vrbit_s8(simde_int8x8_t a) { #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) return vrbit_s8(a); #else return simde_vreinterpret_s8_u8(simde_vrbit_u8(simde_vreinterpret_u8_s8(a))); #endif } #if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) #undef vrbit_s8 #define vrbit_s8(a) simde_vrbit_s8(a) #endif SIMDE_FUNCTION_ATTRIBUTES simde_uint8x16_t simde_vrbitq_u8(simde_uint8x16_t a) { #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) return vrbitq_u8(a); #elif defined(SIMDE_X86_GFNI_NATIVE) return _mm_gf2p8affine_epi64_epi8(a, _mm_set1_epi64x(HEDLEY_STATIC_CAST(int64_t, UINT64_C(0x8040201008040201))), 0); #elif defined(SIMDE_X86_SSE2_NATIVE) __m128i mask; mask = _mm_set1_epi8(0x55); a = _mm_or_si128(_mm_andnot_si128(mask, _mm_slli_epi16(a, 1)), _mm_and_si128(mask, _mm_srli_epi16(a, 1))); mask = _mm_set1_epi8(0x33); a = _mm_or_si128(_mm_andnot_si128(mask, _mm_slli_epi16(a, 2)), _mm_and_si128(mask, _mm_srli_epi16(a, 2))); mask = _mm_set1_epi8(0x0F); a = _mm_or_si128(_mm_andnot_si128(mask, _mm_slli_epi16(a, 4)), _mm_and_si128(mask, _mm_srli_epi16(a, 4))); return a; #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) shift; shift = vec_splat_u8(1); a = vec_sel(vec_sl(a, shift), vec_sr(a, shift), vec_splats(HEDLEY_STATIC_CAST(unsigned char, 0x55))); shift = vec_splat_u8(2); a = vec_sel(vec_sl(a, shift), vec_sr(a, shift), vec_splats(HEDLEY_STATIC_CAST(unsigned char, 0x33))); shift = vec_splat_u8(4); a = vec_or(vec_sl(a, shift), vec_sr(a, shift)); return a; #elif defined(SIMDE_WASM_SIMD128_NATIVE) a = wasm_v128_bitselect(wasm_u8x16_shr(a, 1), wasm_i8x16_shl(a, 1), wasm_i8x16_splat(0x55)); a = wasm_v128_bitselect(wasm_u8x16_shr(a, 2), wasm_i8x16_shl(a, 2), wasm_i8x16_splat(0x33)); a = wasm_v128_or(wasm_u8x16_shr(a, 4), wasm_i8x16_shl(a, 4)); return a; #else simde_uint8x16_private r_, a_ = simde_uint8x16_to_private(a); SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { #if HEDLEY_HAS_BUILTIN(__builtin_bitreverse8) && !defined(HEDLEY_IBM_VERSION) r_.values[i] = __builtin_bitreverse8(a_.values[i]); #else r_.values[i] = HEDLEY_STATIC_CAST(uint8_t, (((a_.values[i] * UINT64_C(0x80200802)) & UINT64_C(0x0884422110)) * UINT64_C(0x0101010101)) >> 32); #endif } return simde_uint8x16_from_private(r_); #endif } #if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) #undef vrbitq_u8 #define vrbitq_u8(a) simde_vrbitq_u8(a) #endif SIMDE_FUNCTION_ATTRIBUTES simde_int8x16_t simde_vrbitq_s8(simde_int8x16_t a) { #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) return vrbitq_s8(a); #else return simde_vreinterpretq_s8_u8(simde_vrbitq_u8(simde_vreinterpretq_u8_s8(a))); #endif } #if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) #undef vrbitq_s8 #define vrbitq_s8(a) simde_vrbitq_s8(a) #endif SIMDE_END_DECLS_ HEDLEY_DIAGNOSTIC_POP #endif /* !defined(SIMDE_ARM_NEON_RBIT_H) */