* 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 sort.h
* \brief
*/
#if !defined(__ASCENDC_INCLUDE_INTERNAL_HEADERS__)
#define __ASCENDC_INCLUDE_INTERNAL_HEADERS__
#define __UNDEF_ASCENDC_INCLUDE_INTERNAL_HEADERS_SORT_H__
#endif
#ifndef LIB_SORT_SORT_H
#define LIB_SORT_SORT_H
#include "kernel_basic_intf.h"
#include "kernel_tensor.h"
#include "include/adv_api/sort/sort_utils.h"
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 3510 || __NPU_ARCH__ == 5102 || __NPU_ARCH__ == 3003 || \
__NPU_ARCH__ == 3113)
#include "../../../impl/adv_api/detail/sort/sort/sort_impl.h"
#endif
* @ingroup Sort
* @brief Sort them according to the value
* @param [out] dstLocal output LocalTensor
* @param [in] concatLocal input LocalTensor
* @param [in] indexLocal input LocalTensor
* @param [in] tmpLocal tmp buffer
* @param [in] repeatTimes repeat times
*
* template <typename T, bool isFullSort>
* __aicore__ inline void Sort(const LocalTensor<T>& dstLocal, const LocalTensor<T>& concatLocal,
* const LocalTensor<uint32_t>& indexLocal, LocalTensor<T>& tmpLocal, const int32_t repeatTimes);
*/
namespace AscendC {
#pragma begin_pipe(V)
* \ingroup Sort
* \brief Sort the given sources or indices accordingly with first calCount elements
* \tparam T: half/float/bfloat16_t/int16_t/uint16_t/int32_t/uint32_t
* \tparam U: uint32_t/int32_t/uint64_t/int64_t
* \tparam isReuseSource: whether allows API to modify source data, usually for performance or
* save less temp bufffer size.
* \tparam config: sort relative configuration.
* \param [out] dstTensor: output LocalTensor for value elements
* \param [out] dstIndexTensor: output LocalTensor for index elements
* \param [in] srcTensor: input LocalTensor for value elements
* \param [in] srcIndexTensor: input LocalTensor for index elements
* \param [in] calCount: the number of elements to be processed.
* \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 acos_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 Tensors must be 32B align, and it doesn't allow address overlap among tensors.
*/
template <typename T, typename U, bool isReuseSource = false, const SortConfig& config = DEFAULT_SORT_CONFIG>
__aicore__ inline void Sort(LocalTensor<T>& dstTensor, LocalTensor<U>& dstIndexTensor, const LocalTensor<T>& srcTensor,
const LocalTensor<U>& srcIndexTensor, const LocalTensor<uint8_t>& sharedTmpBuffer, const uint32_t calCount)
{
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 3510 || __NPU_ARCH__ == 5102 || __NPU_ARCH__ == 3003 || \
__NPU_ARCH__ == 3113)
Reg::internal::SortImpl<T, U, isReuseSource, config>(
dstTensor, dstIndexTensor, srcTensor, srcIndexTensor, sharedTmpBuffer, calCount);
#endif
}
* \ingroup Sort
* \brief Sort the given sources or indices accordingly with first calCount elements
* \tparam T: half/float/bfloat16_t/int16_t/uint16_t/int32_t/uint32_t
* \tparam U: uint32_t/int32_t/uint64_t/int64_t
* \tparam isReuseSource: whether allows API to modify source data, usually for performance or
* save less temp bufffer size.
* \tparam config: sort relative configuration.
* \param [out] dstTensor: output LocalTensor for value elements
* \param [out] dstIndexTensor: output LocalTensor for index elements
* \param [in] srcTensor: input LocalTensor for value elements
* \param [in] srcIndexTensor: input LocalTensor for index elements
* \param [in] calCount: the number of elements to be processed.
* \note Tensors must be 32B align, and it doesn't allow address overlap among tensors.
*/
template <typename T, typename U, bool isReuseSource = false, const SortConfig& config = DEFAULT_SORT_CONFIG>
__aicore__ inline void Sort(const LocalTensor<T>& dstTensor, const LocalTensor<U>& dstIndexTensor,
const LocalTensor<T>& srcTensor, const LocalTensor<U>& srcIndexTensor, const uint32_t calCount)
{
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 3510 || __NPU_ARCH__ == 5102 || __NPU_ARCH__ == 3003 || \
__NPU_ARCH__ == 3113)
Reg::internal::SortImpl<T, U, isReuseSource, config>(
dstTensor, dstIndexTensor, srcTensor, srcIndexTensor, calCount);
#endif
}
* \ingroup Sort
* \brief Sort the given sources accordingly with first calCount elements
* \tparam T: half/float/bfloat16_t/int16_t/uint16_t/int32_t/uint32_t
* \tparam isReuseSource: whether allows API to modify source data, usually for performance or
* save less temp bufffer size.
* \tparam config: sort relative configuration.
* \param [out] dstTensor: output LocalTensor for value elements
* \param [in] srcTensor: input LocalTensor for value elements
* \param [in] calCount: the number of elements to be processed.
* \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 acos_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 Tensors must be 32B align, and it doesn't allow address overlap among tensors.
*/
template <typename T, bool isReuseSource = false, const SortConfig& config = DEFAULT_SORT_CONFIG>
__aicore__ inline void Sort(LocalTensor<T>& dstTensor, const LocalTensor<T>& srcTensor,
const LocalTensor<uint8_t>& sharedTmpBuffer, const uint32_t calCount)
{
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 3510 || __NPU_ARCH__ == 5102 || __NPU_ARCH__ == 3003 || \
__NPU_ARCH__ == 3113)
Reg::internal::SortImpl<T, isReuseSource, config>(dstTensor, srcTensor, sharedTmpBuffer, calCount);
#endif
}
* \ingroup Sort
* \brief Sort the given sources accordingly with first calCount elements
* \tparam T: half/float/bfloat16_t/int16_t/uint16_t/int32_t/uint32_t
* \tparam isReuseSource: whether allows API to modify source data, usually for performance or
* save less temp bufffer size.
* \tparam config: sort relative configuration.
* \param [out] dstTensor: output LocalTensor for value elements
* \param [in] srcTensor: input LocalTensor for value elements
* \param [in] calCount: the number of elements to be processed.
* \note Tensors must be 32B align, and it doesn't allow address overlap among tensors.
*/
template <typename T, bool isReuseSource = false, const SortConfig& config = DEFAULT_SORT_CONFIG>
__aicore__ inline void Sort(LocalTensor<T>& dstTensor, const LocalTensor<T>& srcTensor, const uint32_t calCount)
{
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 3510 || __NPU_ARCH__ == 5102 || __NPU_ARCH__ == 3003 || \
__NPU_ARCH__ == 3113)
Reg::internal::SortImpl<T, isReuseSource, config>(dstTensor, srcTensor, calCount);
#endif
}
* \ingroup Sort
* \brief Sort the given sources accordingly with first calCount elements
* \tparam T: half/float/bfloat16_t/int16_t/uint16_t/int32_t/uint32_t
* \tparam isReuseSource: whether allows API to modify source data, usually for performance or
* save less temp bufffer size.
* \tparam config: sort relative configuration.
* \param [out] dstTensor: output LocalTensor for value elements
* \param [out] dstIndexTensor: output LocalTensor for index elements based on default index for inputs.
* \param [in] srcTensor: input LocalTensor for value elements
* \param [in] calCount: the number of elements to be processed.
* \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 acos_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 Tensors must be 32B align, and it doesn't allow address overlap among tensors.
*/
template <typename T, bool isReuseSource = false, const SortConfig& config = DEFAULT_SORT_CONFIG>
__aicore__ inline void Sort(LocalTensor<T>& dstTensor, LocalTensor<uint32_t>& dstIndexTensor,
const LocalTensor<T>& srcTensor, const LocalTensor<uint8_t>& sharedTmpBuffer, const uint32_t calCount)
{
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 3510 || __NPU_ARCH__ == 5102 || __NPU_ARCH__ == 3003 || \
__NPU_ARCH__ == 3113)
Reg::internal::SortImpl<T, isReuseSource, config>(
dstTensor, dstIndexTensor, srcTensor, sharedTmpBuffer, calCount);
#endif
}
* \ingroup Sort
* \brief Sort the given sources accordingly with first calCount elements
* \tparam T: half/float/bfloat16_t/int16_t/uint16_t/int32_t/uint32_t
* \tparam isReuseSource: whether allows API to modify source data, usually for performance reason or
* save less temp bufffer size.
* \tparam config: sort relative configuration.
* \param [out] dstTensor: output LocalTensor for value elements
* \param [out] dstIndexTensor: output LocalTensor for index elements based on default index for inputs.
* \param [in] srcTensor: input LocalTensor for value elements
* \param [in] calCount: the number of elements to be processed.
* \note Tensors must be 32B align, and it doesn't allow address overlap among tensors.
*/
template <typename T, bool isReuseSource = false, const SortConfig& config = DEFAULT_SORT_CONFIG>
__aicore__ inline void Sort(LocalTensor<T>& dstTensor, LocalTensor<uint32_t>& dstIndexTensor,
const LocalTensor<T>& srcTensor, const uint32_t calCount)
{
#if defined(__NPU_ARCH__) && (__NPU_ARCH__ == 3510 || __NPU_ARCH__ == 5102 || __NPU_ARCH__ == 3003 || \
__NPU_ARCH__ == 3113)
Reg::internal::SortImpl<T, isReuseSource, config>(dstTensor, dstIndexTensor, srcTensor, calCount);
#endif
}
#pragma end_pipe
}
#endif
#if defined(__UNDEF_ASCENDC_INCLUDE_INTERNAL_HEADERS_SORT_H__)
#undef __ASCENDC_INCLUDE_INTERNAL_HEADERS__
#undef __UNDEF_ASCENDC_INCLUDE_INTERNAL_HEADERS_SORT_H__
#endif
