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_GATHER_MESH_1D_DETOUR_H_
12 : #define HCCLV2_CCU_INSTRUCTION_ALL_GATHER_MESH_1D_DETOUR_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 : class CcuCtxArgAllGatherMeshDetour1D : public CcuCtxArg {
25 : public:
26 0 : explicit CcuCtxArgAllGatherMeshDetour1D(const std::vector<uint64_t> &dSize, uint32_t rId, const CollAlgOperator &op,
27 : const std::vector<std::vector<RankId>> &tempVTopo, uint64_t singleTransportSize, uint64_t detourPathNum,
28 0 : uint64_t pathNumPerPeer) :
29 0 : dimSize_(dSize), rankId_(rId), op_(op), tempVTopo_(tempVTopo), singleTransportSize_(singleTransportSize),
30 0 : detourPathNum_(detourPathNum), pathNumPerPeer_(pathNumPerPeer) {}
31 :
32 0 : CcuCtxSignature GetCtxSignature() const override
33 : {
34 0 : CcuCtxSignature signature;
35 0 : GenerateCcuCtxSignature(signature, CcuInstType::CCU_ALLGATHER_MESH_1D_DETOUR, op_, tempVTopo_);
36 0 : return signature;
37 0 : }
38 :
39 : std::vector<uint64_t> dimSize_;
40 : uint32_t rankId_;
41 : CollAlgOperator op_;
42 : std::vector<std::vector<RankId>> tempVTopo_;
43 : uint64_t singleTransportSize_;
44 : uint64_t detourPathNum_;
45 : uint64_t pathNumPerPeer_;
46 : };
47 :
48 : class CcuTaskArgAllGatherMeshDetour1D : public CcuTaskArg {
49 : public:
50 0 : explicit CcuTaskArgAllGatherMeshDetour1D(uint64_t inputAddr, uint64_t outputAddr, uint64_t token,
51 : uint64_t baseOffset, uint64_t tailOffset, uint64_t tailSize, uint64_t loopIterNum,
52 0 : const std::vector<uint64_t> &lengths) :
53 0 : inputAddr_(inputAddr), outputAddr_(outputAddr), token_(token), baseOffset_(baseOffset), tailOffset_(tailOffset),
54 0 : tailSize_(tailSize), loopIterNum_(loopIterNum), lengths_(lengths) {}
55 :
56 : uint64_t inputAddr_;
57 : uint64_t outputAddr_;
58 : uint64_t token_;
59 : uint64_t baseOffset_;
60 : uint64_t tailOffset_;
61 : uint64_t tailSize_; // 尾块数据量
62 : uint64_t loopIterNum_;
63 : std::vector<uint64_t> lengths_; // 每个loop迭代一次搬运时每个ms上的数据量
64 : };
65 :
66 : class CcuInstructionAllGatherMeshDetour1D : public CcuInstruction {
67 : public:
68 0 : CcuInstructionAllGatherMeshDetour1D() : CcuInstruction()
69 : {
70 0 : }
71 :
72 0 : void InitDetourInfo(uint32_t rankId, uint64_t inputAddr, uint64_t outputAddr, uint64_t token, uint64_t baseOffset,
73 : uint64_t tailOffset, uint64_t tailSize, uint64_t loopIterNum, std::vector<uint64_t> &lengths,
74 : uint64_t singleTransportSize, uint64_t detourPathNum, uint64_t pathNumPerPeer, CollAlgOperator &op,
75 : std::vector<std::vector<RankId>> &tempVTopo)
76 : {
77 0 : u32 maxDimNum = 1;
78 0 : if (tempVTopo.size() != maxDimNum) {
79 0 : THROW<InvalidParamsException>(StringFormat(
80 : "[CcuInstructionAllGatherMeshDetour1D] tempVTopo size is not 1, size is [%zu].", tempVTopo.size()));
81 : }
82 0 : dimSize_.push_back(tempVTopo[0].size());
83 0 : rankId_ = rankId;
84 0 : inputAddr_ = inputAddr;
85 0 : outputAddr_ = outputAddr;
86 0 : token_ = token;
87 0 : baseOffset_ = baseOffset;
88 0 : tailOffset_ = tailOffset;
89 0 : tailSize_ = tailSize;
90 0 : loopIterNum_ = loopIterNum;
91 0 : lengths_ = lengths;
92 0 : singleTransportSize_ = singleTransportSize;
93 0 : detourPathNum_ = detourPathNum;
94 0 : pathNumPerPeer_ = pathNumPerPeer;
95 0 : op_ = op;
96 0 : tempVTopo_ = tempVTopo;
97 0 : return;
98 : }
99 :
100 0 : CcuInstType GetInstType() const override
101 : {
102 0 : HCCL_INFO("CcuInstructionAllGatherMeshDetour1D instype is CCU_ALLGATHER_MESH_1D_DETOUR.");
103 0 : return instType_;
104 : }
105 :
106 0 : std::string Describe() const override
107 : {
108 : return StringFormat("CcuInstructionAllGatherMeshDetour1D rankId [%u], instType[%s]",
109 0 : rankId_, instType_.Describe().c_str());
110 : }
111 :
112 0 : std::unique_ptr<CcuCtxArg> GetCtxArg() const override
113 : {
114 0 : return std::make_unique<CcuCtxArgAllGatherMeshDetour1D>(
115 0 : dimSize_, rankId_, op_, tempVTopo_, singleTransportSize_, detourPathNum_, pathNumPerPeer_);
116 : }
117 :
118 0 : std::unique_ptr<CcuTaskArg> GetTaskArg() const override
119 : {
120 0 : return std::make_unique<CcuTaskArgAllGatherMeshDetour1D>(inputAddr_, outputAddr_, token_, baseOffset_,
121 0 : tailOffset_, tailSize_, loopIterNum_, lengths_);
122 : }
123 :
124 : void SetInstType(CcuInstType instType)
125 : {
126 : instType_ = instType;
127 : }
128 : private:
129 : CcuInstType instType_ = CcuInstType::CCU_ALLGATHER_MESH_1D_DETOUR;
130 : std::vector<uint64_t> dimSize_;
131 : uint32_t rankId_{0};
132 : uint64_t inputAddr_{0};
133 : uint64_t outputAddr_{0};
134 : uint64_t token_{0};
135 : uint64_t baseOffset_{0};
136 : uint64_t tailOffset_{0};
137 : uint64_t tailSize_{0};
138 : uint64_t loopIterNum_{0};
139 : std::vector<uint64_t> lengths_;
140 :
141 : uint64_t singleTransportSize_{0};
142 : uint64_t detourPathNum_{0}; // 到每个对端有几个绕路路径
143 : uint64_t pathNumPerPeer_{0};
144 : CollAlgOperator op_;
145 : std::vector<std::vector<RankId>> tempVTopo_;
146 : };
147 :
148 : }
149 : #endif // HCCLV2_CCU_INSTRUCTION_ALL_GATHER_MESH_1D_H_
|
