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              : #include "topo_match_concurr_mesh.h"
      12              : 
      13              : namespace Hccl {
      14            2 : TopoMatchConcurrMesh::TopoMatchConcurrMesh(const RankId vRank, const u32 rankSize, const RankGraph *rankGraph,
      15            2 :                                            const DevType devType)
      16            2 :     : TopoMatchBase(vRank, rankSize, rankGraph, devType)
      17              : {
      18            2 : }
      19              : 
      20            2 : TopoMatchConcurrMesh::~TopoMatchConcurrMesh()
      21              : {
      22            2 : }
      23              : 
      24            2 : HcclResult TopoMatchConcurrMesh::MatchTopo(std::vector<std::vector<RankId>> &vTopo,
      25              :                                            std::vector<RankId> &virtRanks, std::map<RankId, u32> &virtRankMap)
      26              : {
      27              :     // 获取并校验当前通信层数
      28            2 :     std::set<u32> levelSet = rankGraph_->GetLevels(myRank_);
      29              : 
      30            2 :     CHK_PRT_RET((levelSet.size() == COMM_LEVEL_SIZE_0),
      31              :         HCCL_ERROR("[CollAlgFactory] [TopoMatchConcurrMesh] Rank [%d], Invalid virtual topo.", myRank_),
      32              :         HcclResult::HCCL_E_PARA);
      33            2 :     const NetInstance* netInstance = rankGraph_->GetNetInstanceByRankId(0, myRank_);
      34            2 :     std::set<RankId> rankSet = netInstance->GetRankIds();
      35           10 :     for (RankId rankId : rankSet) {
      36            8 :         rankIds_.push_back(rankId);
      37              :     }
      38              :     // 判断level0上的拓扑是否符合 m x n 要求
      39            2 :     CHK_RET(CalcRankOnSamePlaneOfR0(rankOnSameBoardVector_, rankOnSameSlotVector_, numRanksPerBoard_));
      40              : 
      41              :     const auto minmxPair =
      42            2 :         std::minmax_element(numRanksPerBoard_.begin(), numRanksPerBoard_.end());
      43            2 :     u32 minNumRankPerBoard = *minmxPair.first;
      44            2 :     u32 maxNumRankPerBoard = *minmxPair.second;
      45            2 :     CHK_PRT_RET((minNumRankPerBoard != maxNumRankPerBoard),
      46              :                 HCCL_ERROR("[CollAlgFactory] [TopoMatchConcurrMesh] Rank [%d], Invalid virtual topo for "
      47              :                            "multi-dimensional concurrent mesh, min numRanksPerBoard_[%u], max numRanksPerBoard_[%u].",
      48              :                            myRank_, minNumRankPerBoard, maxNumRankPerBoard),
      49              :                 HcclResult::HCCL_E_PARA);
      50              : 
      51            2 :     CHK_PRT_RET(((rankSize_ == 1) || (numRanksPerBoard_[0] * numRanksPerBoard_.size() != rankSize_)),
      52              :                 HCCL_ERROR("[CollAlgFactory] [TopoMatchConcurrMesh] Rank [%d], Invalid virtual topo for "
      53              :                            "multi-dimensional concurrent mesh "
      54              :                            "algorithm with rankSize [%u], ranksPerBoard [%u], ranksPerSlot [%u].",
      55              :                            myRank_, rankSize_, numRanksPerBoard_[0], numRanksPerBoard_.size()),
      56              :                 HcclResult::HCCL_E_PARA);
      57              : 
      58              :     // 计算当前rankd的virtRanks, vTopo和virtRankMap
      59            2 :     u32 myLocalId = rankGraph_->GetReplacedLocalId(myRank_);
      60            2 :     rankOnSameBoard_ = rankOnSameBoardVector_[myLocalId / RANK_SIZE_EIGHT];
      61            2 :     rankOnSameSlot_ = rankOnSameSlotVector_[myLocalId % RANK_SIZE_EIGHT];
      62              : 
      63            2 :     if ((rankOnSameBoard_.size() == 1) || (rankOnSameSlot_.size() == 1)) {
      64            0 :         HCCL_DEBUG("[CollAlgFactory] [TopoMatchConcurrMesh] Rank [%d],Virtual topo with rankSize [%u], ranksPerBoard "
      65              :                    "[%u], ranksPerSlot [%u]. "
      66              :                    "1-D Mesh algorithm is adopted.",
      67              :                    myRank_, rankSize_, rankOnSameBoard_.size(), rankOnSameSlot_.size());
      68              :     }
      69            2 :     sort(rankOnSameBoard_.begin(), rankOnSameBoard_.end());
      70            2 :     sort(rankOnSameSlot_.begin(), rankOnSameSlot_.end());
      71            2 :     vTopo.push_back(rankOnSameBoard_);
      72            2 :     vTopo.push_back(rankOnSameSlot_);
      73              : 
      74            2 :     sort(rankIds_.begin(), rankIds_.end());
      75            2 :     virtRanks = rankIds_;
      76              : 
      77            2 :     CHK_PRT_RET(
      78              :         GenVirtRankMapping(virtRanks, virtRankMap) != HcclResult::HCCL_SUCCESS,
      79              :         HCCL_ERROR("[CollAlgFactory] [TopoMatchConcurrMesh] Rank [%d], Fail to generate virtRankMapping.", myRank_),
      80              :         HcclResult::HCCL_E_INTERNAL);
      81              : 
      82            2 :     return HcclResult::HCCL_SUCCESS;
      83            2 : }
      84              : 
      85              : 
      86              : } // namespace Hccl