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 HCCLV2_CCU_INSTRUCTION_ALL_REDUCE_MESH_1D_ONE_SHOT_H_
      12              : #define HCCLV2_CCU_INSTRUCTION_ALL_REDUCE_MESH_1D_ONE_SHOT_H_
      13              : 
      14              : #include "template_utils.h"
      15              : #include "instruction.h"
      16              : #include "ins_queue.h"
      17              : #include "ccu_context_utils.h"
      18              : #include "ccu_ctx_signature.h"
      19              : #include "ccu_ins.h"
      20              : #include "ccu_rank_group.h"
      21              : 
      22              : namespace Hccl {
      23              : 
      24              : // 为AllReduceMesh1D实现的CCUIns、CCUCtxArg与CCUTaskArg
      25              : class CcuCtxArgAllReduceMesh1DOneShot : public CcuCtxArg {
      26              : public:
      27            0 :     explicit CcuCtxArgAllReduceMesh1DOneShot(const std::vector<uint64_t> &dSize, uint32_t rId,
      28              :                                              const std::string &notifySignal, const CollAlgOperator &op,
      29              :                                              const std::vector<std::vector<RankId>> &tempVTopo)
      30            0 :         : dimSize_(dSize), rankId_(rId), notifySignal_(notifySignal), op_(op), tempVTopo_(tempVTopo)
      31              :     {
      32            0 :     }
      33            0 :     CcuCtxSignature GetCtxSignature() const override
      34              :     {
      35            0 :         CcuCtxSignature signature;
      36            0 :         GenerateCcuCtxSignature(signature, CcuInstType::CCU_ALL_REDUCE_MESH_1D_ONE_SHOT_DIRECT, op_, tempVTopo_);
      37            0 :         return signature;
      38            0 :     }
      39              :     std::vector<uint64_t>            dimSize_;
      40              :     uint32_t                         rankId_;
      41              :     std::string                      notifySignal_;
      42              :     CollAlgOperator                  op_;
      43              :     std::vector<std::vector<RankId>> tempVTopo_;
      44              : };
      45              : 
      46              : class CcuTaskArgAllReduceMesh1DOneShot : public CcuTaskArg {
      47              : public:
      48            0 :     explicit CcuTaskArgAllReduceMesh1DOneShot(uint64_t inputAddr, uint64_t outputAddr, uint64_t sliceSize,
      49            0 :         uint64_t token) :
      50            0 :         inputAddr_(inputAddr), outputAddr_(outputAddr), sliceSize_(sliceSize), token_(token) {}
      51              : 
      52              :     uint64_t inputAddr_;
      53              :     uint64_t outputAddr_;
      54              :     uint64_t sliceSize_;
      55              :     uint64_t token_;
      56              : };
      57              : 
      58              : class CcuInstructionAllReduceMesh1DOneShot : public CcuInstruction {
      59              : public:
      60            0 :     CcuInstructionAllReduceMesh1DOneShot() : CcuInstruction()
      61              :     {
      62            0 :     }
      63              : 
      64            0 :     void Init(uint32_t rankId, uint64_t inputAddr, uint64_t outputAddr, uint64_t sliceSize,
      65              :         uint64_t token, const std::string &notifySignal, CollAlgOperator &op, std::vector<std::vector<RankId>> &tempVTopo)
      66              :     {
      67            0 :         u32 maxDimNum = 1;
      68            0 :         if (tempVTopo.size() != maxDimNum) {
      69            0 :             THROW<InvalidParamsException>(StringFormat(
      70              :                 "[CcuInstructionAllReduceMesh1DOneShot] tempVTopo size is not 1, size is [%zu].", tempVTopo.size()));
      71              :         }
      72            0 :         dimSize_.push_back(tempVTopo[0].size());
      73            0 :         rankId_ = rankId;
      74            0 :         inputAddr_ = inputAddr;
      75            0 :         outputAddr_ = outputAddr;
      76            0 :         sliceSize_ = sliceSize;
      77            0 :         token_ = token;
      78            0 :         op_ = op;
      79            0 :         tempVTopo_ = tempVTopo;
      80            0 :         notifySignal_ = notifySignal;
      81            0 :         return;
      82              :     }
      83              : 
      84            0 :     CcuInstType GetInstType() const override
      85              :     {
      86            0 :         HCCL_INFO("CcuInstructionAllReduceMesh1DOneShot instype is CCU_ALL_REDUCE_MESH_1D_ONE_SHOT_DIRECT.");
      87            0 :         return instType_;
      88              :     }
      89              : 
      90            0 :     std::string Describe() const override
      91              :     {
      92            0 :         return StringFormat("CcuInstructionAllReduceMesh1DOneShot rankId [%u], instType[%s]", rankId_, instType_.Describe().c_str());
      93              :     }
      94              : 
      95            0 :     std::unique_ptr<CcuCtxArg> GetCtxArg() const override
      96              :     {
      97            0 :         return std::make_unique<CcuCtxArgAllReduceMesh1DOneShot>(dimSize_, rankId_, notifySignal_, op_, tempVTopo_);
      98              :     }
      99              : 
     100              :     void SetInstType(CcuInstType instType) 
     101              :     { 
     102              :         instType_ = instType; 
     103              :     }
     104              : 
     105            0 :     std::unique_ptr<CcuTaskArg> GetTaskArg() const override
     106              :     {
     107            0 :         return std::make_unique<CcuTaskArgAllReduceMesh1DOneShot>(inputAddr_, outputAddr_, sliceSize_, token_);
     108              :     }
     109              : 
     110              : private:
     111              :     CcuInstType instType_ = CcuInstType::CCU_ALL_REDUCE_MESH_1D_ONE_SHOT_DIRECT;
     112              :     std::string notifySignal_ = "Reduce_defalut";
     113              :     std::vector<uint64_t> dimSize_;
     114              :     uint32_t rankId_{0};
     115              :     uint64_t inputAddr_{0};
     116              :     uint64_t outputAddr_{0};
     117              :     uint64_t sliceSize_{0};
     118              :     uint64_t token_{0};
     119              :     CollAlgOperator op_;
     120              :     std::vector<std::vector<RankId>> tempVTopo_;
     121              : };
     122              : 
     123              : }
     124              : #endif // HCCLV2_CCU_INSTRUCTION_ALL_REDUCE_MESH_1D_H_