/* 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: * 2023 Yi-Yen Chung (Copyright owned by Andes Technology) */ #if !defined(SIMDE_ARM_NEON_QRDMLAH_H) #define SIMDE_ARM_NEON_QRDMLAH_H #include "types.h" #include "qmovn.h" HEDLEY_DIAGNOSTIC_PUSH SIMDE_DISABLE_UNWANTED_DIAGNOSTICS SIMDE_BEGIN_DECLS_ SIMDE_FUNCTION_ATTRIBUTES int16_t simde_vqrdmlahh_s16(int16_t a, int16_t b, int16_t c) { #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) && defined(SIMDE_ARCH_ARM_QRDMX) #if defined(__clang__) && !SIMDE_DETECT_CLANG_VERSION_CHECK(11,0,0) return SIMDE_DISABLE_DIAGNOSTIC_EXPR_(SIMDE_DIAGNOSTIC_DISABLE_VECTOR_CONVERSION_, vqrdmlahh_s16(a, b, c)); #else return vqrdmlahh_s16(a, b, c); #endif #else int64_t r = (((1 << 15) + (HEDLEY_STATIC_CAST(int64_t, a) << 16) + ((HEDLEY_STATIC_CAST(int64_t, (HEDLEY_STATIC_CAST(int64_t, b) * HEDLEY_STATIC_CAST(int64_t, c)))) << 1)) >> 16); return simde_vqmovns_s32(HEDLEY_STATIC_CAST(int32_t, r)); #endif } #if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) #undef vqrdmlahh_s16 #define vqrdmlahh_s16(a, b, c) simde_vqrdmlahh_s16((a), (b), (c)) #endif SIMDE_FUNCTION_ATTRIBUTES int32_t simde_vqrdmlahs_s32(int32_t a, int32_t b, int32_t c) { #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) && defined(SIMDE_ARCH_ARM_QRDMX) return vqrdmlahs_s32(a, b, c); #else int64_t round_const = (HEDLEY_STATIC_CAST(int64_t, 1) << 31); int64_t a_ = (HEDLEY_STATIC_CAST(int64_t, a) << 32); int64_t sum = round_const + a_; int64_t mul = (HEDLEY_STATIC_CAST(int64_t, b) * HEDLEY_STATIC_CAST(int64_t, c)); int64_t mul2 = mul << 1; if (mul2 >> 1 != mul) { if (mul > 0) return INT32_MAX; else if (mul < 0) return INT32_MIN; } int64_t sum2 = sum + mul2; if (sum > 0 && INT64_MAX - sum < mul2) return INT32_MAX; if (sum < 0 && INT64_MIN - sum > mul2) return INT32_MIN; return HEDLEY_STATIC_CAST(int32_t, ((sum2 >> 32) & 0xffffffff)); #endif } #if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) #undef vqrdmlahs_s32 #define vqrdmlahs_s32(a, b, c) simde_vqrdmlahs_s32((a), (b), (c)) #endif SIMDE_FUNCTION_ATTRIBUTES simde_int16x4_t simde_vqrdmlah_s16(simde_int16x4_t a, simde_int16x4_t b, simde_int16x4_t c) { #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) && defined(SIMDE_ARCH_ARM_QRDMX) return vqrdmlah_s16(a, b, c); #else simde_int16x4_private r_, a_ = simde_int16x4_to_private(a), b_ = simde_int16x4_to_private(b), c_ = simde_int16x4_to_private(c); SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { r_.values[i] = simde_vqrdmlahh_s16(a_.values[i], b_.values[i], c_.values[i]); } return simde_int16x4_from_private(r_); #endif } #if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) #undef vqrdmlah_s16 #define vqrdmlah_s16(a, b, c) simde_vqrdmlah_s16((a), (b), (c)) #endif SIMDE_FUNCTION_ATTRIBUTES simde_int32x2_t simde_vqrdmlah_s32(simde_int32x2_t a, simde_int32x2_t b, simde_int32x2_t c) { #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) && defined(SIMDE_ARCH_ARM_QRDMX) return vqrdmlah_s32(a, b, c); #else simde_int32x2_private r_, a_ = simde_int32x2_to_private(a), b_ = simde_int32x2_to_private(b), c_ = simde_int32x2_to_private(c); SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { r_.values[i] = simde_vqrdmlahs_s32(a_.values[i], b_.values[i], c_.values[i]); } return simde_int32x2_from_private(r_); #endif } #if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) #undef vqrdmlah_s32 #define vqrdmlah_s32(a, b, c) simde_vqrdmlah_s32((a), (b), (c)) #endif SIMDE_FUNCTION_ATTRIBUTES simde_int16x8_t simde_vqrdmlahq_s16(simde_int16x8_t a, simde_int16x8_t b, simde_int16x8_t c) { #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) && defined(SIMDE_ARCH_ARM_QRDMX) return vqrdmlahq_s16(a, b, c); #else simde_int16x8_private r_, a_ = simde_int16x8_to_private(a), b_ = simde_int16x8_to_private(b), c_ = simde_int16x8_to_private(c); SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { r_.values[i] = simde_vqrdmlahh_s16(a_.values[i], b_.values[i], c_.values[i]); } return simde_int16x8_from_private(r_); #endif } #if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) #undef vqrdmlahq_s16 #define vqrdmlahq_s16(a, b, c) simde_vqrdmlahq_s16((a), (b), (c)) #endif SIMDE_FUNCTION_ATTRIBUTES simde_int32x4_t simde_vqrdmlahq_s32(simde_int32x4_t a, simde_int32x4_t b, simde_int32x4_t c) { #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) && defined(SIMDE_ARCH_ARM_QRDMX) return vqrdmlahq_s32(a, b, c); #else simde_int32x4_private r_, a_ = simde_int32x4_to_private(a), b_ = simde_int32x4_to_private(b), c_ = simde_int32x4_to_private(c); SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) { r_.values[i] = simde_vqrdmlahs_s32(a_.values[i], b_.values[i], c_.values[i]); } return simde_int32x4_from_private(r_); #endif } #if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) #undef vqrdmlahq_s32 #define vqrdmlahq_s32(a, b, c) simde_vqrdmlahq_s32((a), (b), (c)) #endif SIMDE_END_DECLS_ HEDLEY_DIAGNOSTIC_POP #endif /* !defined(SIMDE_ARM_NEON_QRDMLAH_H) */