Line data    Source code

       1              : /**
       2              :  * Copyright (c) 2025 Huawei Technologies Co., Ltd.
       3              :  * This program is free software, you can redistribute it and/or modify it under the terms and conditions of
       4              :  * CANN Open Software License Agreement Version 2.0 (the "License").
       5              :  * Please refer to the License for details. You may not use this file except in compliance with the License.
       6              :  * THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
       7              :  * INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
       8              :  * See LICENSE in the root of the software repository for the full text of the License.
       9              :  */
      10              : 
      11              : #ifndef OPERATOR_KERNEL_POST_PROCESS_DYNAMIC_OUTPUT_H
      12              : #define OPERATOR_KERNEL_POST_PROCESS_DYNAMIC_OUTPUT_H
      13              : 
      14              : #include <vector>
      15              : #include "operator_kernel.h"
      16              : 
      17              : 
      18              : namespace AicpuSchedule {
      19              : class OperatorKernelPostProcessDynamicOutput : public OperatorKernel {
      20              : public:
      21           16 :     OperatorKernelPostProcessDynamicOutput() = default;
      22           22 :     ~OperatorKernelPostProcessDynamicOutput() = default;
      23              :     int32_t Compute(const AicpuTaskInfo &kernelTaskInfo, const RunContext &taskContext) override;
      24              : private:
      25              :     int32_t CopyTensorDesc(const PostprocessDynamicOutputKernelArgs * const param, const RunContext &taskContext) const;
      26              :     int32_t AllocatedOutput(const PostprocessDynamicOutputKernelArgs * const param, Mbuf *const outputMbuf,
      27              :                             const uint32_t dynamicFlag, const size_t index, RuntimeTensorDesc **dynamicSrcDescPptr,
      28              :                             RuntimeTensorDesc **staticSrcDescPptr) const;
      29              :     int32_t GetRuntimeTensor(const PostprocessDynamicOutputKernelArgs * const param,
      30              :                              RuntimeTensorDesc **dynamicSrcDescPtr) const;
      31              :     int32_t PostprocessSetDataLen(const RuntimeTensorDesc * const runtimeTensorDesc, Mbuf *const outputMbuf,
      32              :                                   const size_t index) const;
      33              :     int32_t GetMbufHeadFromResp(Mbuf **const respMbufPtr, const RunContext &taskContext, void **const customBufPtr,
      34              :                                 uint32_t *const customBufSizePtr) const;
      35              :     int32_t PostProcessForOutputToAllocate(const PostprocessDynamicOutputKernelArgs * const param,
      36              :                                            Mbuf **const outputMbufPtr, const size_t index,
      37              :                                            RuntimeTensorDesc **dynamicSrcDescPptr, void *const customBuf,
      38              :                                            const uint32_t customBufSize, const RunContext &taskContext) const;
      39              :     int32_t MakeupDynamicOutputMbuf(Mbuf **const outputMbufPtr, const size_t index,
      40              :                                     const RuntimeTensorDesc *const runtimeTensorDesc, void *const customBuf,
      41              :                                     const uint32_t customBufSize, const RunContext &taskContext) const;
      42              :     int32_t FreeMbuf(const PostprocessDynamicOutputKernelArgs * const param, const RunContext &taskContext) const;
      43              : 
      44              :     bool IsSupportSdmaCopy() const;
      45              : 
      46              :     bool OptimizedMemCopy(void *const dst_data, const size_t dst_size,
      47              :                           const void *const src_data, const size_t src_size) const;
      48              : };
      49              : }  // namespace AicpuSchedule
      50              : #endif  // OPERATOR_KERNEL_POST_PROCESS_DYNAMIC_OUTPUT_H