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 "coll_executor_base.h"
12 : #include "hccl_aiv.h"
13 :
14 : namespace hccl {
15 :
16 181 : CollExecutorBase::CollExecutorBase(const HcclDispatcher dispatcher, std::unique_ptr<TopoMatcher> &topoMatcher)
17 181 : : dispatcher_(dispatcher), topoMatcher_(topoMatcher)
18 : {
19 168 : }
20 :
21 97 : HcclResult CollExecutorBase::SetAlgType(const AlgType algType)
22 : {
23 97 : const std::vector<AlgTypeLevel0> &l0Algo = desc_.level0SupportedAlgos;
24 97 : const std::vector<AlgTypeLevel1> &l1Algo = desc_.level1SupportedAlgos;
25 97 : const std::vector<AlgTypeLevel2> &l2Algo = desc_.level2SupportedAlgos;
26 :
27 97 : algType_ = algType;
28 97 : if (!l0Algo.empty()
29 94 : && std::find(l0Algo.begin(), l0Algo.end(), algType_.algoLevel0) == l0Algo.end()) {
30 0 : HCCL_WARNING("[%s] not support level0 algo[%d], reset to algo[%d]", __func__,
31 : algType_.algoLevel0, l0Algo[0]);
32 0 : algType_.algoLevel0 = l0Algo[0];
33 : }
34 94 : if (!l1Algo.empty()
35 95 : && std::find(l1Algo.begin(), l1Algo.end(), algType_.algoLevel1) == l1Algo.end()) {
36 0 : HCCL_WARNING("[%s] not support level1 algo[%d], reset to algo[%d]", __func__,
37 : algType_.algoLevel1, l1Algo[0]);
38 0 : algType_.algoLevel1 = l1Algo[0];
39 : }
40 95 : if (!l2Algo.empty()
41 93 : && std::find(l2Algo.begin(), l2Algo.end(), algType_.algoLevel2) == l2Algo.end()) {
42 1 : HCCL_WARNING("[%s] not support level2 algo[%d], reset to algo[%d]", __func__,
43 : algType_.algoLevel2, l2Algo[0]);
44 1 : algType_.algoLevel2 = l2Algo[0];
45 : }
46 : // 记录刷新后的算法类型到executor的描述中
47 93 : desc_.algType = algType_;
48 93 : return HCCL_SUCCESS;
49 : }
50 :
51 75 : HcclResult CollExecutorBase::SetCCLInBuffer(u64 cclbufferSize)
52 : {
53 75 : inCCLbufferSize_ = cclbufferSize;
54 75 : return HCCL_SUCCESS;
55 : }
56 :
57 35 : HcclResult CollExecutorBase::SetIsSupportSDMAReduce(bool isSupportSDMAReduce)
58 : {
59 35 : isSupportSDMAReduce_ = isSupportSDMAReduce;
60 35 : return HCCL_SUCCESS;
61 : }
62 :
63 0 : HcclResult CollExecutorBase::RunTemplate(const std::unique_ptr<AlgTemplateBase> &tempAlg, const SubCommInfo &commInfo)
64 : {
65 0 : HcclResult ret = tempAlg->RunAsync(commInfo.localRank, commInfo.localRankSize, commInfo.links);
66 0 : CHK_PRT_RET(ret == HCCL_E_AGAIN, HCCL_WARNING("[CollExecutorBase][RunTemplate]" \
67 : "group has been destroyed. Break!"), ret);
68 0 : CHK_PRT_RET(ret != HCCL_SUCCESS,
69 : HCCL_ERROR("[CollExecutorBase][RunTemplate]run tempAlg rank[%u] rank size[%u] failed",
70 : commInfo.localRank, commInfo.localRankSize), ret);
71 0 : return HCCL_SUCCESS;
72 : }
73 :
74 1 : HcclResult CollExecutorBase::GetAivExecParam(const OpParam& param, AlgResourceResponse& algRes, AivSuperKernelArgs &args)
75 : {
76 1 : return HCCL_SUCCESS;
77 : }
78 :
79 25 : HcclResult CollExecutorBase::PrepareCommInfoToDevice(AlgResourceResponse& algResource)
80 : {
81 : (void) algResource;
82 25 : return HCCL_SUCCESS;
83 : }
84 :
85 0 : HcclResult CollExecutorBase::SetRmaInfo(void* rmaInfo)
86 : {
87 0 : CHK_PTR_NULL(rmaInfo);
88 0 : rmaInfo_ = rmaInfo;
89 0 : return HCCL_SUCCESS;
90 : }
91 :
92 0 : HcclResult CollExecutorBase::CalcIncreLinkRequest(const OpParam& param, std::set<u32>& ranksLinked,
93 : AlgResourceRequest &resourceRequest, bool& needIncreLink)
94 : {
95 : (void) ranksLinked;
96 : (void) needIncreLink;
97 0 : return HCCL_SUCCESS;
98 : }
99 :
100 0 : HcclResult CollExecutorBase::CreatePairWiseList(HcclSendRecvItem *sendRecvInfo, u32 itemNum)
101 : {
102 0 : return HCCL_SUCCESS;
103 : }
104 :
105 0 : HcclResult CollExecutorBase::GetPairWiseList(std::vector<std::vector<HcclSendRecvItem*>> &sendRecvPairList)
106 : {
107 0 : return HCCL_SUCCESS;
108 : }
109 :
110 24 : HcclResult CollExecutorBase::SetAlgOpContext(AlgOpContext algOpContext)
111 : {
112 24 : algOpContext_ = algOpContext;
113 24 : return HCCL_SUCCESS;
114 : }
115 :
116 24 : HcclResult CollExecutorBase::SetAivClearEnable(bool aivClearEnable)
117 : {
118 24 : aivClearEnable_ = aivClearEnable;
119 24 : return HCCL_SUCCESS;
120 : }
121 :
122 1 : HcclResult CollExecutorBase::CalNumBlocks(u32& numBlocks, u32 rankSize, u64 dataSize, HcclCMDType cmdType)
123 : {
124 1 : numBlocks = rankSize;
125 1 : return HCCL_SUCCESS;
126 : }
127 :
128 21 : HcclResult CollExecutorBase::GetNumBlocks(u32& numBlocks) const
129 : {
130 21 : numBlocks = numBlocks_;
131 21 : return HCCL_SUCCESS;
132 : }
133 :
134 6 : HcclResult CollExecutorBase::SetNumBlocks(const u32& numBlocks)
135 : {
136 6 : numBlocks_ = numBlocks;
137 6 : return HCCL_SUCCESS;
138 : }
139 :
140 2 : HcclResult CollExecutorBase::GetCache(HcclCacheInfo& cacheInfo){
141 2 : cacheInfo = cacheInfo_;
142 2 : return HCCL_SUCCESS;
143 : }
144 :
145 24 : HcclResult CollExecutorBase::SetOpCounter(const OpCounterInfo& opCounter)
146 : {
147 24 : opCounter_ = opCounter;
148 24 : return HCCL_SUCCESS;
149 : }
150 0 : HcclResult CollExecutorBase::GetAdjInfo(AlgResourceResponse& algRes, AdjInfo& adjInfo)
151 : {
152 : (void) adjInfo;
153 0 : return HCCL_SUCCESS;
154 : }
155 :
156 0 : HcclResult CollExecutorBase::MarkNeedAlltoallvCache()
157 : {
158 0 : HCCL_ERROR("[CollExecutorBase][MarkNeedAlltoallvCache] not supported for current executor!");
159 0 : return HCCL_E_NOT_SUPPORT;
160 : }
161 :
162 0 : HcclResult CollExecutorBase::GetHcclOffsetDstRanksMap(std::unordered_map<uint64_t, std::vector<uint32_t>>& hcclOffsetDstRanksMap) const
163 : {
164 : UNUSED_PARAM(hcclOffsetDstRanksMap);
165 0 : HCCL_ERROR("[CollExecutorBase][GetHcclOffsetDstRanksMap] not supported for current executor!");
166 0 : return HCCL_E_NOT_SUPPORT;
167 : }
168 :
169 0 : void CollExecutorBase::SetWorkflowMode(HcclWorkflowMode workflowMode)
170 : {
171 : UNUSED_PARAM(workflowMode);
172 0 : HCCL_DEBUG("[CollExecutorBase][SetWorkflowMode] base class default, no effect");
173 0 : }
174 :
175 0 : HcclWorkflowMode CollExecutorBase::GetExecutorWorkflowMode() const
176 : {
177 0 : HCCL_ERROR("[CollExecutorBase][GetExecutorWorkflowMode] base class should not be called");
178 0 : return HcclWorkflowMode::HCCL_WORKFLOW_MODE_OP_BASE;
179 : }
180 :
181 : }
|
