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_PREPARE_OUTPUT_H
      12              : #define OPERATOR_KERNEL_PREPARE_OUTPUT_H
      13              : 
      14              : #include "operator_kernel.h"
      15              : #include "operator_kernel_context.h"
      16              : 
      17              : 
      18              : namespace AicpuSchedule {
      19              : class OperatorKernelPrepareOutputBase {
      20              : public:
      21           72 :     OperatorKernelPrepareOutputBase() = default;
      22           72 :     virtual ~OperatorKernelPrepareOutputBase() {};
      23              : 
      24              :     int32_t PrepareOutput(ProcessOutputInfo &outputInfo, const RunContext &taskContext, const bool zeroCpy,
      25              :                           RuntimeTensorDesc *const tensorDesc) const;
      26              :     int32_t PrepareOutWithTensorDesc(const AicpuTaskInfo &kernelTaskInfo, const bool zeroCpy,
      27              :                                      const RunContext &taskContext) const;
      28              : private:
      29              :     int32_t PrepareOutputNonZeroCpy(const ProcessOutputInfo &outputInfo, Mbuf * const outMBuf,
      30              :                                     RuntimeTensorDesc *const tensorDesc) const;
      31              :     int32_t GetStaticNNOutPutIndex(const uint32_t modelId) const;
      32              :     void MarkStaticNNOutPutIndex(const uint32_t modelId) const;
      33              : };
      34              : 
      35              : class OperatorKernelModelPrepareOutput : public OperatorKernel, public OperatorKernelPrepareOutputBase {
      36              : public:
      37           15 :     OperatorKernelModelPrepareOutput() = default;
      38           18 :     ~OperatorKernelModelPrepareOutput() = default;
      39              : 
      40              :     int32_t Compute(const AicpuTaskInfo &kernelTaskInfo, const RunContext &taskContext) override;
      41              : };
      42              : 
      43              : class OperatorKernelModelPrepareOutputWithTensorDesc : public OperatorKernel, public OperatorKernelPrepareOutputBase {
      44              : public:
      45           15 :     OperatorKernelModelPrepareOutputWithTensorDesc() = default;
      46           18 :     ~OperatorKernelModelPrepareOutputWithTensorDesc() = default;
      47              : 
      48              :     int32_t Compute(const AicpuTaskInfo &kernelTaskInfo, const RunContext &taskContext) override;
      49              : };
      50              : 
      51              : class OperatorKernelBufferPrepareOutput : public OperatorKernel, public OperatorKernelPrepareOutputBase {
      52              : public:
      53           15 :     OperatorKernelBufferPrepareOutput() = default;
      54           18 :     ~OperatorKernelBufferPrepareOutput() = default;
      55              : 
      56              :     int32_t Compute(const AicpuTaskInfo &kernelTaskInfo, const RunContext &taskContext) override;
      57              : };
      58              : 
      59              : class OperatorKernelBufferPrepareOutputWithTensorDesc : public OperatorKernel, public OperatorKernelPrepareOutputBase {
      60              : public:
      61           15 :     OperatorKernelBufferPrepareOutputWithTensorDesc() = default;
      62           18 :     ~OperatorKernelBufferPrepareOutputWithTensorDesc() = default;
      63              : 
      64              :     int32_t Compute(const AicpuTaskInfo &kernelTaskInfo, const RunContext &taskContext) override;
      65              : };
      66              : 
      67              : }  // namespace AicpuSchedule
      68              : 
      69              : #endif  // OPERATOR_KERNEL_PREPARE_OUTPUT_H