* 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 kernel_operator_vec_duplicate_intf_impl.h
* \brief
*/
#if !defined(__ASCENDC_INCLUDE_INTERNAL_HEADERS__)
#pragma message("impl/basic_api/kernel_operator_vec_duplicate_intf_impl.h is an internal header file and must not be used directly. Functions or variables defined in this file may be removed in the future. Please use \"#include \"basic_api/kernel_operator_vec_duplicate_intf.h\"\" and use public functions or variables defined in interface headers files.")
#define __ASCENDC_INCLUDE_INTERNAL_HEADERS__
#define __UNDEF_ASCENDC_INCLUDE_INTERNAL_HEADERS_KERNEL_OPERATOR_VEC_DUPLICATE_INTF_IMPL_H__
#endif
#ifndef ASCENDC_MODULE_OPERATOR_VEC_DUPLICATE_INTERFACE_IMPL_H
#define ASCENDC_MODULE_OPERATOR_VEC_DUPLICATE_INTERFACE_IMPL_H
#include "kernel_tensor.h"
#include "kernel_check.h"
#include "kernel_npu_debug.h"
#include "mstx_local_tensor_info.h"
#if __NPU_ARCH__ == 1001
#include "dav_c100/kernel_operator_vec_duplicate_impl.h"
#elif __NPU_ARCH__ == 2002
#include "dav_m200/kernel_operator_vec_duplicate_impl.h"
#elif __NPU_ARCH__ == 2201
#include "dav_c220/kernel_operator_vec_duplicate_impl.h"
#elif __NPU_ARCH__ == 3002
#include "dav_m300/kernel_operator_vec_duplicate_impl.h"
#elif __NPU_ARCH__ == 3102
#include "dav_m310/kernel_operator_vec_duplicate_impl.h"
#elif __NPU_ARCH__ == 3510
#include "reg_compute/kernel_reg_compute_intf.h"
#include "dav_3510/kernel_operator_vec_duplicate_impl.h"
#elif (__NPU_ARCH__ == 5102)
#include "reg_compute/kernel_reg_compute_intf.h"
#include "dav_m510/kernel_operator_vec_duplicate_impl.h"
#elif __NPU_ARCH__ == 3003
#include "dav_l300/kernel_operator_vec_duplicate_impl.h"
#elif __NPU_ARCH__ == 3113
#include "dav_l311/kernel_operator_vec_duplicate_impl.h"
#endif
#pragma begin_pipe(V)
namespace AscendC {
* Duplicate *
* ************************************************************************************************* */
* @ingroup Duplicate Level 0
* @brief dst[i] = scalar
* @param [out] dst output LocalTensor
* @param [in] scalar input scalar number
* @param [in] mask[]/mask mask array/count
* @param [in] repeatTime repeat times
* @param [in] dstBlockStride dst block stride
* @param [in] dstRepeatStride dst repeat stride
*/
template <typename T, bool isSetMask>
__aicore__ inline void Duplicate(const LocalTensor<T>& dst, const T& scalarValue, uint64_t mask,
const uint8_t repeatTime, const uint16_t dstBlockStride, const uint8_t dstRepeatStride)
{
#if defined(ASCENDC_DEBUG) || defined(ASCENDC_CPU_DEBUG)
CheckVectorTensor("Duplicate", NamedTensor(dst, "dst"));
CheckMaskRepeat<T, isSetMask>(mask, repeatTime, "Duplicate");
#endif
#ifdef __MSTX_DFX_REPORT__
MstxTensor::GetMstxVecDupInfo(dst, mask, repeatTime, dstBlockStride, dstRepeatStride, isSetMask, "Duplicate");
#endif
CheckDuplicateSupportedType<T>();
#if ASCENDC_CPU_DEBUG
MaskSetter::Instance().SetMask(isSetMask);
if (!CheckFunDup(dst, mask, repeatTime, dstBlockStride, dstRepeatStride, "Duplicate")) {
ASCENDC_REPORT_CHECK_ERROR("Duplicate", KernelFuncType::MASK_COUNT_MODE);
}
#endif
DuplicateImpl<T, isSetMask>((__ubuf__ T*)dst.GetPhyAddr(), scalarValue, mask, repeatTime, dstBlockStride,
dstRepeatStride);
}
template <typename T, bool isSetMask>
__aicore__ inline void Duplicate(const LocalTensor<T>& dst, const T& scalarValue, uint64_t mask[],
const uint8_t repeatTime, const uint16_t dstBlockStride, const uint8_t dstRepeatStride)
{
#if defined(ASCENDC_DEBUG) || defined(ASCENDC_CPU_DEBUG)
CheckVectorTensor("Duplicate", NamedTensor(dst, "dst"));
CheckMaskRepeat<T, isSetMask>(mask, repeatTime, "Duplicate");
#endif
#ifdef __MSTX_DFX_REPORT__
MstxTensor::GetMstxVecDupInfo(dst, mask[0], mask[1], repeatTime, dstBlockStride, dstRepeatStride, isSetMask, "Duplicate");
#endif
CheckDuplicateSupportedType<T>();
#if ASCENDC_CPU_DEBUG
MaskSetter::Instance().SetMask(isSetMask);
if (!CheckFunDup(dst, mask, repeatTime, dstBlockStride, dstRepeatStride, "Duplicate")) {
ASCENDC_REPORT_CHECK_ERROR("Duplicate", KernelFuncType::MASK_BIT_MODE);
}
#endif
DuplicateImpl<T, isSetMask>((__ubuf__ T*)dst.GetPhyAddr(), scalarValue, mask, repeatTime, dstBlockStride,
dstRepeatStride);
}
* @ingroup Duplicate Level 2
* @brief dst = dst[i] = scalar
* @param [out] dst output LocalTensor
* @param [in] scalar input scalar number
* @param [in] count number Number of data involved in calculation
*/
template <typename T>
__aicore__ inline void Duplicate(const LocalTensor<T>& dst, const T& scalarValue, const int32_t& count)
{
#if defined(ASCENDC_DEBUG) || defined(ASCENDC_CPU_DEBUG)
CheckVectorTensor("Duplicate", NamedTensor(dst, "dst"));
CheckCalcount(count, "count", "Duplicate");
#endif
#ifdef __MSTX_DFX_REPORT__
MstxTensor::GetMstxVecDupInfo(dst, count, "Duplicate");
#endif
CheckDuplicateSupportedType<T>();
#if ASCENDC_CPU_DEBUG
if (!CheckFunDup(dst, count, "Duplicate")) {
ASCENDC_REPORT_CHECK_ERROR("Duplicate", KernelFuncType::CALCOUNT_MODE);
}
#endif
DuplicateImpl<T>((__ubuf__ T*)dst.GetPhyAddr(), scalarValue, count);
}
#if (__NPU_ARCH__ == 3510) || (__NPU_ARCH__ == 5102) || (__NPU_ARCH__ == 3003) || (__NPU_ARCH__ == 3113)
* @ingroup Duplicate lowest position of src
* @brief dst = dst[i] = src[0]
* @param [out] dst output LocalTensor
* @param [in] src input LocalTensor
* @param [in] count number Number of data involved in calculation
*/
template <typename T>
__aicore__ inline void Duplicate(const LocalTensor<T>& dst, const LocalTensor<T>& src,
const int32_t& count)
{
#ifdef __MSTX_DFX_REPORT__
MstxTensor::GetMstxVecDupInfo(dst, count, "Duplicate");
#endif
using PrimType = PrimT<T>;
CheckDuplicateSupportedType<PrimType>();
#if ASCENDC_CPU_DEBUG
if (!CheckFunDup(dst, count, "Duplicate")) {
ASCENDC_REPORT_CHECK_ERROR("Duplicate", KernelFuncType::CALCOUNT_MODE);
}
#endif
DuplicateImpl<PrimType>((__ubuf__ PrimType*)dst.GetPhyAddr(), (__ubuf__ PrimType*)src.GetPhyAddr(), count);
}
* Interleave *
* ************************************************************************************************* */
* @ingroup Interleave Level 2
* @brief Interleave src0 and src1 to dst0 and dst1
* @param [out] dst0 output0 LocalTensor
* @param [out] dst1 output1 LocalTensor
* @param [in] src0 input0 LocalTensor
* @param [in] src1 input1 LocalTensor
* @param [count] count number of data calculation, must be even number
*/
template <typename T>
__aicore__ inline void Interleave(const LocalTensor<T> &dst0, const LocalTensor<T> &dst1,
const LocalTensor<T> &src0, const LocalTensor<T> &src1, const int32_t count)
{
using PrimType = PrimT<T>;
InterleaveImpl<PrimType>((__ubuf__ PrimType *)dst0.GetPhyAddr(),
(__ubuf__ PrimType *)dst1.GetPhyAddr(),
(__ubuf__ PrimType *)src0.GetPhyAddr(),
(__ubuf__ PrimType *)src1.GetPhyAddr(),
count);
}
* DeInterleave *
* ************************************************************************************************* */
* @ingroup DeInterleave Level 2
* @brief DeInterleave src0 and src1 to dst0 and dst1
* @param [out] dst0 output0 LocalTensor
* @param [out] dst1 output1 LocalTensor
* @param [in] src0 input0 LocalTensor
* @param [in] src1 input1 LocalTensor
* @param [count] count number of data calculation, must be even number
*/
template <typename T>
__aicore__ inline void DeInterleave(const LocalTensor<T> &dst0, const LocalTensor<T> &dst1,
const LocalTensor<T> &src0, const LocalTensor<T> &src1, const int32_t count)
{
using PrimType = PrimT<T>;
DeInterleaveImpl<PrimType>((__ubuf__ PrimType *)dst0.GetPhyAddr(),
(__ubuf__ PrimType *)dst1.GetPhyAddr(),
(__ubuf__ PrimType *)src0.GetPhyAddr(),
(__ubuf__ PrimType *)src1.GetPhyAddr(),
count);
}
* @ingroup DeInterleave Level 2
* @brief DeInterleave src to dst0 and dst1
* @param [out] dst0 output0 LocalTensor
* @param [out] dst1 output1 LocalTensor
* @param [in] src input LocalTensor
* @param [srcCount] srcCount number of data calculation, must be even number
*/
template <typename T>
__aicore__ inline void DeInterleave(const LocalTensor<T> &dst0, const LocalTensor<T> &dst1,
const LocalTensor<T> &src, const int32_t srcCount)
{
using PrimType = PrimT<T>;
DeInterleaveImpl<PrimType>((__ubuf__ PrimType *)dst0.GetPhyAddr(),
(__ubuf__ PrimType *)dst1.GetPhyAddr(),
(__ubuf__ PrimType *)src.GetPhyAddr(),
srcCount);
}
#endif
}
#pragma end_pipe
#endif
#if defined(__UNDEF_ASCENDC_INCLUDE_INTERNAL_HEADERS_KERNEL_OPERATOR_VEC_DUPLICATE_INTF_IMPL_H__)
#undef __ASCENDC_INCLUDE_INTERNAL_HEADERS__
#undef __UNDEF_ASCENDC_INCLUDE_INTERNAL_HEADERS_KERNEL_OPERATOR_VEC_DUPLICATE_INTF_IMPL_H__
#endif
