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 CUPT_KERNEL_CCU_CONTEXT_REDUCE_SCATTER_NHR_1D_H_
      12              : #define CUPT_KERNEL_CCU_CONTEXT_REDUCE_SCATTER_NHR_1D_H_
      13              : #include <vector>
      14              : #include <ios>
      15              : #include "log.h"
      16              : #include "ccu_context_alg_base.h"
      17              : #include "ccu_datatype.h"
      18              : #include "ccu_instruction_reduce_scatter_nhr1d_mem2mem.h"
      19              : 
      20              : namespace Hccl {
      21              : 
      22              : class CcuContextReduceScatterNHR1DMem2Mem : public CcuContextAlgBase {
      23              : public:
      24              :     CcuContextReduceScatterNHR1DMem2Mem(const CcuCtxArg &arg, const std::vector<CcuTransport*> &transports,
      25              :                                                    const CcuTransportGroup &group);
      26            0 :     ~CcuContextReduceScatterNHR1DMem2Mem() override {}
      27              : 
      28              :     void Algorithm() override;
      29              :     std::vector<uint64_t> GeneArgs(const CcuTaskArg &arg) override;
      30              : private:
      31              :     void LoadArgs();
      32              :     void InitResources();
      33              :     void PreSync();
      34              :     void PostSync();
      35              :     void AxisSync(uint32_t signalIndex);
      36              :     void DoRepeatReduceScatterNHR();
      37              :     void DoRepeatReduceScatterNHRSingleStep(const NHRStepInfo &nhrStepInfo,
      38              :         const std::vector<CcuRep::Variable> &inputSliceOffset);
      39              :     void DoRepeatSendRecvSlices(const u32 &toRank, CcuRep::Memory &src, CcuRep::Memory &dst);
      40              : 
      41              :     // 构造函数中
      42              :     uint64_t dimSize_{0};
      43              :     uint32_t axisId_{0};
      44              :     uint32_t localSize_{0};  // 本rank所在行或列的总rank数
      45              :     uint32_t myRankIdx_{0};
      46              :     uint32_t signalNum_{0};  // 需要使用的signal数量
      47              :     DataType dataType_;
      48              :     DataType outputDataType_;
      49              :     std::vector<NHRStepInfo> stepInfoVector_;   // nhr算法执行过程中的参数
      50              :     std::map<u32, u32> indexMap_;
      51              :     uint32_t       linkNum_{0};
      52              : 
      53              :     // load进来参数
      54              :     std::vector<CcuRep::Variable> input_;
      55              :     CcuRep::Variable output_;
      56              :     std::vector<CcuRep::Variable> token_;
      57              :     CcuRep::Variable die0Size_;
      58              :     CcuRep::Variable die1Size_;
      59              :     CcuRep::Variable inputSliceStride_;
      60              :     CcuRep::Variable outputSliceStride_;
      61              :     CcuRep::Variable inputRepeatStride_;
      62              :     CcuRep::Variable outputRepeatStride_;
      63              :     CcuRep::Variable repeatNumVar_;
      64              :     CcuRep::Variable isBottom_;
      65              :     CcuRep::Variable repeatNumVarTemp_;
      66              :     // 用于记录每次writereduce的数据
      67              :     CcuRep::Variable sliceSize_;
      68              : 
      69              :     // 跨轴同步信号
      70              :     std::string        localAxisSignalName_;
      71              :     std::string        anotherAxisSignalName_;
      72              :     CcuRep::MaskSignal localAxisSignal_;
      73              :     CcuRep::MaskSignal anotherAxisSignal_;
      74              :     CcuRep::MaskSignal localSignal_;
      75              : 
      76              :     CcuRep::Variable repeatInputOffset_;
      77              :     CcuRep::Variable repeatOutputOffset_;
      78              :     CcuRep::Variable myrankInputSliceOffset_;
      79              : 
      80              :     CcuRep::Memory srcMem_;
      81              :     CcuRep::Memory dstMem_;
      82              :     CcuRep::Variable flag_; // 用于判断是否是第一次循环
      83              : };
      84              : } // namespace Hccl
      85              : 
      86              : #endif // HCCLV2_CCU_CONTEXT_REDUCE_SCATTER_NHR_1D_H_