* Copyright (c) 2025 Huawei Technologies Co., Ltd.
* This program is free software, you can redistribute it and/or modify it under the terms and conditions of
* CANN Open Software License Agreement Version 2.0 (the "License").
* Please refer to the License for details. You may not use this file except in compliance with the License.
* THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
* INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
* See LICENSE in the root of the software repository for the full text of the License.
*/
* \file erfc.h
* \brief Defines a series of interface used to do elementwise math Erfc calculation.
* Formula: Supplement of Gauss error function. Erfc(x) = 1 - Erf(x)
* The Erfc function does not have an elementary function expression, and there is calculating by
* function approximation.
* The approximate calculation formula is as follows:
* Erfc(x) = (-xa^2)*(R(z)/S(z))*(x/xa)+�?-x/xa)
* xa = |x| + min_float
* z = min(xa, 10)
* min_float is the smallest value could be represented by float.
* R(z) = (((((((z * R0 + R1) * z + R2) * z + R3) * z + R4) * z + R5) * z + R6) * z + R7) * z + R8
* S(z) = (((((z + S1) * z + S2) * z + S3) * z + S4) * z + S5
* R0 = 0.1735313680e-7
* R1 = -0.9856738394e-6
* R2 = 0.2517003236e-4
* R3 = -0.3848015171e-3
* R4 = 0.5681528564e0
* R5 = 0.5245623129e1
* R6 = 0.2107740710e2
* R7 = 0.4212761755e2
* R8 = 0.4380524149e2
* S1 = 0.9349684299e1
* S2 = 0.3756930664e2
* S3 = 0.8058268949e2
* S4 = 0.9155653738e2
* S5 = 0.4380524152e2
*/
#if !defined(__ASCENDC_INCLUDE_INTERNAL_HEADERS__)
#define __ASCENDC_INCLUDE_INTERNAL_HEADERS__
#define __UNDEF_ASCENDC_INCLUDE_INTERNAL_HEADERS_ERFC_H__
#endif
#ifndef LIB_MATH_ERFC_H
#define LIB_MATH_ERFC_H
#include "kernel_tensor.h"
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 2201 || __NPU_ARCH__ == 2002 || __NPU_ARCH__ == 3510 || \
__NPU_ARCH__ == 5102 || __NPU_ARCH__ == 3003 || __NPU_ARCH__ == 3113)
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 3510 || __NPU_ARCH__ == 5102 || __NPU_ARCH__ == 3003 || \
__NPU_ARCH__ == 3113)
#include "../../../impl/adv_api/detail/math/erfc/erfc_3510_impl.h"
#else
#include "../../../impl/adv_api/detail/math/erfc/erfc_common_impl.h"
#endif
#endif
namespace AscendC {
#pragma begin_pipe(V)
* \ingroup Erfc
* \brief compute Erfc elementwisely
* \tparam T: half/float
* \tparam isReuseSource: whether allows API to modify source data, usually for performance reason
* \param [out] dstTensor: output LocalTensor
* \param [in] srcTensor: input LocalTensor
* \param [in] sharedTmpBuffer: extra temporary shared space used for intermediate values among calculation process,
* whose required space size should refer to corresponding tiling API, which is defined at erfc_tiling.h.
* Generally, the more space you allocate, the better performance you will achieve, and the performance
* reaches peak when buffer size is maximum(calculated by tiling function). Moreover, it is not guaranteed
* that the shared space will be cleared after usage, the data could be anything.
* \param [in] calCount: the number of elements to be processed.
* \note src/dst Tensor must be 32B aligned, and it doesn't allow src/dst/sharedTmpBuffer tensor address overlap.
*/
template <typename T, bool isReuseSource = false>
__aicore__ inline void Erfc(const LocalTensor<T>& dstTensor, const LocalTensor<T>& srcTensor,
const LocalTensor<uint8_t>& sharedTmpBuffer, const uint32_t calCount)
{
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 2201 || __NPU_ARCH__ == 2002 || __NPU_ARCH__ == 3510 || \
__NPU_ARCH__ == 5102 || __NPU_ARCH__ == 3003 || __NPU_ARCH__ == 3113)
ErfcImpl<T, isReuseSource>(dstTensor, srcTensor, sharedTmpBuffer, calCount);
#endif
}
* \ingroup Erfc
* \brief compute Erfc elementwisely for whole source tensor
* \tparam T: half/float
* \tparam isReuseSource: whether allows API to modify source data, usually for performance reason
* \param [out] dstTensor: output LocalTensor
* \param [in] srcTensor: input LocalTensor
* \note src/dst Tensor must be 32B aligned, and it doesn't allow src/dst/sharedTmpBuffer tensor address overlap.
*/
template <typename T, bool isReuseSource = false>
__aicore__ inline void Erfc(const LocalTensor<T>& dstTensor, const LocalTensor<T>& srcTensor)
{
Erfc<T, isReuseSource>(dstTensor, srcTensor, srcTensor.GetSize());
}
* \ingroup Erfc
* \brief compute Erfc elementwisely
* \tparam T: half/float
* \tparam isReuseSource: whether allows API to modify source data, usually for performance reason
* \param [out] dstTensor: output LocalTensor
* \param [in] srcTensor: input LocalTensor
* \param [in] calCount: the number of elements to be processed.
* \note src/dst Tensor must be 32B aligned, and it doesn't allow src/dst/sharedTmpBuffer tensor address overlap.
*/
template <typename T, bool isReuseSource = false>
__aicore__ inline void Erfc(const LocalTensor<T>& dstTensor, const LocalTensor<T>& srcTensor, const uint32_t calCount)
{
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 2201 || __NPU_ARCH__ == 2002 || __NPU_ARCH__ == 3510 || \
__NPU_ARCH__ == 5102 || __NPU_ARCH__ == 3003 || __NPU_ARCH__ == 3113)
ErfcImpl<T, isReuseSource>(dstTensor, srcTensor, calCount);
#endif
}
* \ingroup Erfc
* \brief compute Erfc elementwisely for whole source tensor
* \tparam T: half/float
* \tparam isReuseSource: whether allows API to modify source data, usually for performance reason
* \param [out] dstTensor: output LocalTensor
* \param [in] srcTensor: input LocalTensor
* \param [in] sharedTmpBuffer: extra temporary shared space used for intermediate values among calculation process,
* whose required space size should refer to corresponding tiling API, which is defined at erfc_tiling.h.
* Generally, the more space you allocate, the better performance you will achieve, and the performance
* reaches peak when buffer size is maximum(calculated by tiling function). Moreover, it is not guaranteed
* that the shared space will be cleared after usage, the data could be anything.
* \note src/dst Tensor must be 32B aligned, and it doesn't allow src/dst/sharedTmpBuffer tensor address overlap.
*/
template <typename T, bool isReuseSource = false>
__aicore__ inline void Erfc(const LocalTensor<T>& dstTensor, const LocalTensor<T>& srcTensor,
const LocalTensor<uint8_t>& sharedTmpBuffer)
{
Erfc<T, isReuseSource>(dstTensor, srcTensor, sharedTmpBuffer, srcTensor.GetSize());
}
#pragma end_pipe
}
#endif
#if defined(__UNDEF_ASCENDC_INCLUDE_INTERNAL_HEADERS_ERFC_H__)
#undef __ASCENDC_INCLUDE_INTERNAL_HEADERS__
#undef __UNDEF_ASCENDC_INCLUDE_INTERNAL_HEADERS_ERFC_H__
#endif
