diff options
Diffstat (limited to 'third_party/simde/simde/arm/neon/recpe.h')
-rw-r--r-- | third_party/simde/simde/arm/neon/recpe.h | 265 |
1 files changed, 265 insertions, 0 deletions
diff --git a/third_party/simde/simde/arm/neon/recpe.h b/third_party/simde/simde/arm/neon/recpe.h new file mode 100644 index 0000000000..ed9ef42548 --- /dev/null +++ b/third_party/simde/simde/arm/neon/recpe.h @@ -0,0 +1,265 @@ +/* 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> + * 2021 Zhi An Ng <zhin@google.com> (Copyright owned by Google, LLC) + */ + +#if !defined(SIMDE_ARM_NEON_RECPE_H) +#define SIMDE_ARM_NEON_RECPE_H + +#include "types.h" + +HEDLEY_DIAGNOSTIC_PUSH +SIMDE_DISABLE_UNWANTED_DIAGNOSTICS +SIMDE_BEGIN_DECLS_ + +SIMDE_FUNCTION_ATTRIBUTES +simde_float32_t +simde_vrecpes_f32(simde_float32_t a) { + #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) + return vrecpes_f32(a); + #else + return SIMDE_FLOAT32_C(1.0) / a; + #endif +} +#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) + #undef vrecpes_f32 + #define vrecpes_f32(a) simde_vrecpes_f32((a)) +#endif + +SIMDE_FUNCTION_ATTRIBUTES +simde_float64_t +simde_vrecped_f64(simde_float64_t a) { + #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) + return vrecped_f64(a); + #else + return SIMDE_FLOAT64_C(1.0) / a; + #endif +} +#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) + #undef vrecped_f64 + #define vrecped_f64(a) simde_vrecped_f64((a)) +#endif + +SIMDE_FUNCTION_ATTRIBUTES +simde_float32x2_t +simde_vrecpe_f32(simde_float32x2_t a) { + #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) + return vrecpe_f32(a); + #else + simde_float32x2_private + r_, + a_ = simde_float32x2_to_private(a); + + #if defined(SIMDE_IEEE754_STORAGE) + /* https://stackoverflow.com/questions/12227126/division-as-multiply-and-lut-fast-float-division-reciprocal/12228234#12228234 */ + SIMDE_VECTORIZE + for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { + int32_t ix; + simde_float32 fx = a_.values[i]; + simde_memcpy(&ix, &fx, sizeof(ix)); + int32_t x = INT32_C(0x7EF311C3) - ix; + simde_float32 temp; + simde_memcpy(&temp, &x, sizeof(temp)); + r_.values[i] = temp * (SIMDE_FLOAT32_C(2.0) - temp * fx); + } + #elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) + r_.f32 = 1.0f / a_.f32; + #else + SIMDE_VECTORIZE + for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) { + r_.values[i] = simde_vrecpes_f32(a_.values[i]); + } + #endif + + return simde_float32x2_from_private(r_); + #endif +} +#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES) + #undef vrecpe_f32 + #define vrecpe_f32(a) simde_vrecpe_f32((a)) +#endif + +SIMDE_FUNCTION_ATTRIBUTES +simde_float64x1_t +simde_vrecpe_f64(simde_float64x1_t a) { + #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) + return vrecpe_f64(a); + #else + simde_float64x1_private + r_, + a_ = simde_float64x1_to_private(a); + + #if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) + r_.values = 1.0 / a_.values; + #else + SIMDE_VECTORIZE + for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { + r_.values[i] = simde_vrecped_f64(a_.values[i]); + } + #endif + + return simde_float64x1_from_private(r_); + #endif +} +#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) + #undef vrecpe_f64 + #define vrecpe_f64(a) simde_vrecpe_f64((a)) +#endif + +SIMDE_FUNCTION_ATTRIBUTES +simde_float64x2_t +simde_vrecpeq_f64(simde_float64x2_t a) { + #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) + return vrecpeq_f64(a); + #else + simde_float64x2_private + r_, + a_ = simde_float64x2_to_private(a); + + #if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) + r_.values = 1.0 / a_.values; + #else + SIMDE_VECTORIZE + for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { + r_.values[i] = simde_vrecped_f64(a_.values[i]); + } + #endif + + return simde_float64x2_from_private(r_); + #endif +} +#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) + #undef vrecpeq_f64 + #define vrecpeq_f64(a) simde_vrecpeq_f64((a)) +#endif + +SIMDE_FUNCTION_ATTRIBUTES +simde_float32x4_t +simde_vrecpeq_f32(simde_float32x4_t a) { + #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) + return vrecpeq_f32(a); + #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) + return vec_re(a); + #else + simde_float32x4_private + r_, + a_ = simde_float32x4_to_private(a); + + #if defined(SIMDE_X86_SSE_NATIVE) + r_.m128 = _mm_rcp_ps(a_.m128); + #elif defined(SIMDE_IEEE754_STORAGE) + /* https://stackoverflow.com/questions/12227126/division-as-multiply-and-lut-fast-float-division-reciprocal/12228234#12228234 */ + SIMDE_VECTORIZE + for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { + int32_t ix; + simde_float32 fx = a_.values[i]; + simde_memcpy(&ix, &fx, sizeof(ix)); + int32_t x = INT32_C(0x7EF311C3) - ix; + simde_float32 temp; + simde_memcpy(&temp, &x, sizeof(temp)); + r_.values[i] = temp * (SIMDE_FLOAT32_C(2.0) - temp * fx); + } + #elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) + r_.f32 = 1.0f / a_.f32; + #else + SIMDE_VECTORIZE + for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) { + r_.values[i] = simde_vrecpes_f32(a_.values[i]); + } + #endif + + return simde_float32x4_from_private(r_); + #endif +} +#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES) + #undef vrecpeq_f32 + #define vrecpeq_f32(a) simde_vrecpeq_f32((a)) +#endif + +SIMDE_FUNCTION_ATTRIBUTES +simde_uint32x2_t +simde_vrecpe_u32(simde_uint32x2_t a){ + #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) + return vrecpe_u32(a); + #else + simde_uint32x2_private + a_ = simde_uint32x2_to_private(a), + r_; + + SIMDE_VECTORIZE + for(size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { + if(a_.values[i] <= 0x7FFFFFFF){ + r_.values[i] = UINT32_MAX; + } else { + uint32_t a_temp = (a_.values[i] >> 23) & 511; + a_temp = a_temp * 2 + 1; + uint32_t b = (1 << 19) / a_temp; + r_.values[i] = (b+1) / 2; + r_.values[i] = r_.values[i] << 23; + } + } + + return simde_uint32x2_from_private(r_); + #endif +} +#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES) + #undef vrecpe_u32 + #define vrecpe_u32(a) simde_vrecpe_u32((a)) +#endif + +SIMDE_FUNCTION_ATTRIBUTES +simde_uint32x4_t +simde_vrecpeq_u32(simde_uint32x4_t a){ + #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) + return vrecpeq_u32(a); + #else + simde_uint32x4_private + a_ = simde_uint32x4_to_private(a), + r_; + + SIMDE_VECTORIZE + for(size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { + if(a_.values[i] <= 0x7FFFFFFF){ + r_.values[i] = UINT32_MAX; + } else { + uint32_t a_temp = (a_.values[i] >> 23) & 511; + a_temp = a_temp * 2 + 1; + uint32_t b = (1 << 19) / a_temp; + r_.values[i] = (b+1) / 2; + r_.values[i] = r_.values[i] << 23; + } + } + + return simde_uint32x4_from_private(r_); + #endif +} +#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES) + #undef vrecpeq_u32 + #define vrecpeq_u32(a) simde_vrecpeq_u32((a)) +#endif + +SIMDE_END_DECLS_ +HEDLEY_DIAGNOSTIC_POP +#endif /* !defined(SIMDE_ARM_NEON_RECPE_H) */ |