* 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 softmaxgrad.h
* \brief SoftmaxGrad and SoftmaxGradFront api of AscendC
*/
#ifndef LIB_SOFTMAX_SOFTMAXGRAD_H
#define LIB_SOFTMAX_SOFTMAXGRAD_H
#include "kernel_tensor.h"
#include "kernel_tiling/kernel_tiling.h"
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 2201 || __NPU_ARCH__ == 2002 || __NPU_ARCH__ == 3101 || __NPU_ARCH__ == 3002 || __NPU_ARCH__ == 5102 || \
__NPU_ARCH__ == 3003 || __NPU_ARCH__ == 3113)
#include "../../../impl/adv_api/detail/activation/softmax/softmax_common.h"
#include "../../../impl/adv_api/detail/activation/softmax/softmax_grad_base_impl.h"
#pragma begin_pipe(V)
namespace AscendC {
* \ingroup SoftmaxGrad
* \brief compute process: sum = rowsum(grad * x), grad * x - sum * x
* if isFront = true
* y = sum
* if isFront = false
* y = grad * x - sum * x
* \note support data type: half and float
* \param [out] dstTensor: output y
* \param [in] gradTensor: input grad
* \param [in] srcTensor: input x
* \param [in] softmaxShapeInfo: input x shape
* \param [in] tiling: input softmaxtiling
* \param [in] isFront: compute mode refer to brief
* \param [in] isReuseSource: reserved param
* \param [in] isDataFormatNZ: if the data format of input srcTensor is NZ
*/
template <typename T, bool isReuseSource = false, bool isDataFormatNZ = false>
__aicore__ inline void SoftmaxGrad(const LocalTensor<T>& dstTensor, const LocalTensor<T>& gradTensor,
const LocalTensor<T>& srcTensor, const SoftMaxTiling& tiling, bool isFront = false,
const SoftMaxShapeInfo& softmaxShapeInfo = {})
{
if ASCEND_IS_AIC {
return;
}
TRACE_START(TraceId::SoftmaxGrad);
SoftmaxGradImpl<T, isReuseSource, isDataFormatNZ>(dstTensor, gradTensor, srcTensor, tiling, isFront, softmaxShapeInfo);
TRACE_STOP(TraceId::SoftmaxGrad);
}
* \ingroup SoftmaxGradFront
* \brief compute process: y = rowsum(grad * x)
* \note support data type: half and float
* \param [out] dstTensor: output y
* \param [in] gradTensor: input grad
* \param [in] srcTensor: input x
* \param [in] softmaxShapeInfo: input x shape
* \param [in] tiling: input softmaxtiling
* \param [in] isBasicBlock: if src shape[m,k] satisfy the condition(m%8 == 0 && k%64 == 0), you can set true to improve
* performance , but it is a reserved param when isDataFormatNZ = true
* \param [in] isDataFormatNZ: if the data format of input srcTensor is NZ
*/
template <typename T, bool isBasicBlock = false, bool isDataFormatNZ = false>
__aicore__ inline void SoftmaxGradFront(const LocalTensor<T>& dstTensor, const LocalTensor<T>& gradTensor,
const LocalTensor<T>& srcTensor, const SoftMaxTiling& tiling, const SoftMaxShapeInfo& softmaxShapeInfo = {})
{
if ASCEND_IS_AIC {
return;
}
TRACE_START(TraceId::SoftmaxGrad);
SoftmaxGradFrontImpl<T, isBasicBlock, isDataFormatNZ>(dstTensor, gradTensor, srcTensor, tiling, softmaxShapeInfo);
TRACE_STOP(TraceId::SoftmaxGrad);
}
* \ingroup SoftmaxGrad
* \brief compute process: sum = rowsum(grad * x), grad * x - sum * x
* if isFront = true
* y = sum
* if isFront = false
* y = grad * x - sum * x
* \note support data type: half and float
* \param [out] dstTensor: output y
* \param [in] gradTensor: input grad
* \param [in] srcTensor: input x
* \param [in] softmaxShapeInfo: input x shape
* \param [in] sharedTmpBuffer: input local temporary Tensor,you can get the range by tilingfunc of
* GetSoftMaxGradMinTmpSize/GetSoftMaxGradMaxTmpSize
* \param [in] tiling: input softmaxtiling
* \param [in] isFront: compute mode refer to brief
* \param [in] isReuseSource: reserved param
* \param [in] isDataFormatNZ: if the data format of input srcTensor is NZ
*/
template <typename T, bool isReuseSource = false, bool isDataFormatNZ = false>
__aicore__ inline void SoftmaxGrad(const LocalTensor<T>& dstTensor, const LocalTensor<T>& gradTensor,
const LocalTensor<T>& srcTensor, const LocalTensor<uint8_t>& sharedTmpBuffer, const SoftMaxTiling& tiling,
bool isFront = false, const SoftMaxShapeInfo& softmaxShapeInfo = {})
{
if ASCEND_IS_AIC {
return;
}
TRACE_START(TraceId::SoftmaxGrad);
SoftmaxGradImpl<T, isReuseSource, isDataFormatNZ>(dstTensor, gradTensor, srcTensor, sharedTmpBuffer, tiling, isFront,
softmaxShapeInfo);
TRACE_STOP(TraceId::SoftmaxGrad);
}
* \ingroup SoftmaxGradFront
* \brief compute process: y = rowsum(grad * x)
* \note support data type: half and float
* \param [out] dstTensor: output y
* \param [in] gradTensor: input grad
* \param [in] srcTensor: input x
* \param [in] softmaxShapeInfo: input x shape
* \param [in] sharedTmpBuffer: input local temporary Tensor,you can get the range by tilingfunc of
* GetSoftMaxGradMinTmpSize/GetSoftMaxGradMaxTmpSize
* \param [in] tiling: input softmaxtiling
* \param [in] isBasicBlock: if src shape[m,k] satisfy the condition(m%8 == 0 && k%64 == 0), you can set true to improve
* performance , but it is a reserved param when isDataFormatNZ = true
* \param [in] isDataFormatNZ: if the data format of input srcTensor is NZ
*/
template <typename T, bool isBasicBlock = false, bool isDataFormatNZ = false>
__aicore__ inline void SoftmaxGradFront(const LocalTensor<T>& dstTensor, const LocalTensor<T>& gradTensor,
const LocalTensor<T>& srcTensor, const LocalTensor<uint8_t>& sharedTmpBuffer, const SoftMaxTiling& tiling,
const SoftMaxShapeInfo& softmaxShapeInfo = {})
{
if ASCEND_IS_AIC {
return;
}
SoftmaxGradFrontImpl<T, isBasicBlock, isDataFormatNZ>(dstTensor, gradTensor, srcTensor, sharedTmpBuffer, tiling,
softmaxShapeInfo);
}
}
#pragma end_pipe
#endif
#endif
