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 COLL_SERVICE_DEVICE_MODE_H
      12              : #define COLL_SERVICE_DEVICE_MODE_H
      13              : 
      14              : #include <memory>
      15              : #include "coll_service_base.h"
      16              : #include "communicator_impl.h"
      17              : #include "ins_queue.h"
      18              : #include "ccu_communicator.h"
      19              : #include "ccu_ins_preprocessor.h"
      20              : #include "aicpu_ins_preprocessor.h"
      21              : #include "mc2_compont.h"
      22              : #include "aicpu_stream_manager.h"
      23              : #include "aiv_mc2_compont.h"
      24              : #include "aiv_ins_preprocessor.h"
      25              : #include "aiv_ins.h"
      26              : #include "hccl_aiv_utils.h"
      27              : 
      28              : namespace Hccl {
      29              : 
      30              : class CollServiceDeviceMode : public CollServiceBase {
      31              : public:
      32          237 :     explicit CollServiceDeviceMode(CommunicatorImpl *comm)
      33          237 :         : CollServiceBase(comm), ccuInsPreprocessor(comm), aivInsPreprocessor(comm), aicpuInsPreprocessor(comm),
      34          474 :           mc2Compont(comm), aivMc2Compont(comm)
      35              :     {
      36          237 :     }
      37              : 
      38              :     void Init() override;
      39              :     // 单算子
      40              :     void LoadWithOpBasedMode(CollOperator &op, std::unique_ptr<Stream> stream) override;
      41              :     // 图模式
      42              :     void LoadWithOffloadMode(CollOperator &op, std::unique_ptr<Stream> stream) override;
      43              : 
      44              :     // MC2 CCU用
      45              :     void AllocCommResource(void *mc2Tiling, void **commContext, const AcceleratorState& tilingAccelerator) override;
      46              :     void GetCcuTaskInfo(void *tilingData, void *ccuTaskGroup) override;
      47              :     u32  GetCcuMc2ServerNum() override;
      48              :     CcuInsPreprocessor *GetCcuInsPreprocessor();
      49              :     AivInsPreprocessor *GetAivInsPreprocessor();
      50              :     std::vector<LinkData> GetUniqueLinks(std::shared_ptr<InsQueue> &insQueue) const;
      51              :     // Aicpu用
      52              :     AicpuInsPreprocessor *GetAicpuInsPreprocessor();
      53              :     bool       IsAicpuResExisted(std::string algName);
      54              :     DevBuffer *GetAicpuResBuffer(std::string algName);
      55              : 
      56              :     // 快照保存和恢复场景使用
      57              :     void RecoverTransport(vector<LinkData> &links, vector<std::pair<LinkGroup, u32>> linkGroupPair) override;
      58              :     void RecoverAicpuTransport(vector<LinkData> &links) const;
      59              :     void RecoverCcuTransport(vector<LinkData> &links, vector<std::pair<LinkGroup, u32>> linkGroupPair);
      60              :     HcclResult GetSnapShotDynamicBuf(CollOperator &op,BinaryStream &buf) override;
      61              :     bool IsAllTransportRecoveredReady(const std::string &opTag) override;
      62              : 
      63            2 :     const Mc2Compont& GetMc2Compont() const { return mc2Compont; }
      64              :     const AivMc2Compont& GetAivMc2Compont() const { return aivMc2Compont; }
      65              :     // N秒快恢场景使用
      66              :     void Resume() override;
      67              : 
      68              :     HcclResult GetAlgExecParam(bool clearEnable, u32 numBlocks, void *&commContext, u64 &len) override;
      69              : private:
      70              :     CcuInsPreprocessor   ccuInsPreprocessor;
      71              :     AivInsPreprocessor   aivInsPreprocessor;
      72              :     AicpuInsPreprocessor aicpuInsPreprocessor;
      73              :     AicpuStreamManager   aicpuStreamManager;
      74              : 
      75              :     Mc2Compont mc2Compont;
      76              :     AivMc2Compont aivMc2Compont;
      77              :     std::unordered_set<u32> captureModelIds;
      78              : 
      79              :     std::shared_ptr<InsQueue> Orchestrate(const CollAlgOperator &op) const;
      80              : 
      81              :     // aicpu展开模式单算子快照恢复使用
      82              :     void RecoverInterRankNotifies(const vector<LinkData> &links) const;
      83              : 
      84              :     // AIV场景Acl Graph专用
      85              :     HcclResult HandleAclGraphFirstOpAivBuff(rtStream_t mainStream);
      86              :     // 生成AivOpArgs，AIV superKernel
      87              :     HcclResult GenerateAivOpArgs(const AivInstruction &aivInstruction, AivOpArgs& aivOpArgs) const;
      88              :     void GeneratorAivSuperKernelArgs(const AivOpArgs &aivOpArgs, bool clearEnable, u32 numBlocks,
      89              :                                      AivSuperKernelArgs &superArgs) const;
      90              : };
      91              : 
      92              : } // namespace Hccl
      93              : 
      94              : #endif // COLL_SERVICE_DEVICE_MODE_H