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 "scatter_auto_selector.h"
12 : #include "selector_registry.h"
13 : #include "coll_operator.h"
14 :
15 : namespace Hccl {
16 0 : SelectorStatus ScatterAutoSelector::SelectCcuMsAlgo(const TopoInfo &topoInfo, const CollAlgOperator &op,
17 : const std::map<OpType, std::vector<HcclAlgoType>> &configAlgMap, std::string &primQueueGenName) const
18 : {
19 : (void)topoInfo;
20 : (void)op;
21 : (void)configAlgMap;
22 : (void)primQueueGenName;
23 0 : HCCL_WARNING("[Algo][ScatterAutoSelector] not supported yet for ccu_ms mode, reset to default.");
24 0 : return SelectorStatus::NOT_MATCH;
25 : }
26 :
27 0 : SelectorStatus ScatterAutoSelector::SelectCcuScheduleAlgo(const TopoInfo &topoInfo, const CollAlgOperator &op,
28 : const std::map<OpType, std::vector<HcclAlgoType>> &configAlgMap, std::string &primQueueGenName) const
29 : {
30 : (void)op;
31 : (void)configAlgMap;
32 0 : HCCL_DEBUG("[ScatterAutoSelector][%s] start, topoInfo levelNum[%u]", __func__, topoInfo.levelNum);
33 :
34 0 : if (topoInfo.levelNum > 1) {
35 0 : if (topoInfo.level0Shape == Level0Shape::MESH_1D) {
36 0 : if (topoInfo.netLayerDetails.localNetInsSizeOfLayer[0] == 1) {
37 : // 每框出 1 卡
38 0 : primQueueGenName = "CcuScatterNHRMem2Mem1D";
39 0 : } else if (Is2DieFullMesh()) {
40 0 : HCCL_WARNING("[Algo][ScatterAutoSelector] 2DieFullMesh is not supported yet for schedule mode.");
41 0 : return SelectorStatus::NOT_MATCH;
42 : } else {
43 0 : primQueueGenName = "CcuScatterParallelMesh1DNHR";
44 : }
45 : } else {
46 0 : HCCL_WARNING("[Algo][ScatterAutoSelector] level0Shape[%d] is not supported yet for ccu schedule mode.",
47 : topoInfo.level0Shape);
48 0 : return SelectorStatus::NOT_MATCH;
49 : }
50 : } else {
51 0 : if (topoInfo.level0Shape == Level0Shape::MESH_1D) {
52 0 : if (Is2DieFullMesh()) {
53 0 : HCCL_WARNING("[Algo][ScatterAutoSelector] 2DieFullMesh is not supported yet for ccu schedule mode.");
54 0 : return SelectorStatus::NOT_MATCH;
55 : } else {
56 0 : primQueueGenName = "CcuScatterMesh1D";
57 : }
58 0 : } else if (topoInfo.level0Shape == Level0Shape::MESH_2D) {
59 0 : primQueueGenName = "CcuScatterMesh2D";
60 0 : } else if (topoInfo.level0Shape == Level0Shape::MESH_1D_CLOS) {
61 0 : if (IsLayerAllConnetedWithTopo(topoInfo, 0, TopoType::MESH_1D)) {
62 : // MESH_1D 即可链接所有卡， 使用 MESH_1D 算法
63 0 : primQueueGenName = "CcuScatterMesh1D";
64 0 : } else if (topoInfo.level0PcieMix) {
65 0 : HCCL_WARNING("[Algo][ScatterAutoSelector] level0 PCIE mix is not supported yet for ccu schedule mode.");
66 0 : return SelectorStatus::NOT_MATCH;
67 : } else {
68 0 : primQueueGenName = "CcuAllGatherParallelMesh1DNHR";
69 : }
70 0 : } else if (topoInfo.level0Shape == Level0Shape::CLOS) {
71 0 : HCCL_WARNING("[Algo][ScatterAutoSelector] level0Shape[%d] is not supported yet for ccu schedule mode.",
72 : topoInfo.level0Shape);
73 0 : return SelectorStatus::NOT_MATCH;
74 : } else {
75 0 : HCCL_WARNING("[Algo][ScatterAutoSelector] level0Shape[%d] is not supported yet for ccu schedule mode.",
76 : topoInfo.level0Shape);
77 0 : return SelectorStatus::NOT_MATCH;
78 : }
79 : }
80 0 : HCCL_INFO("[Algo][ScatterAutoSelector][%s] Algo match [%s]", __func__, primQueueGenName.c_str());
81 0 : return SelectorStatus::MATCH;
82 : }
83 :
84 0 : SelectorStatus ScatterAutoSelector::SelectAicpuAlgo(const TopoInfo &topoInfo, const CollAlgOperator &op,
85 : const std::map<OpType, std::vector<HcclAlgoType>> &configAlgMap, std::string &primQueueGenName) const
86 : {
87 : (void)op;
88 : (void)configAlgMap;
89 0 : HCCL_DEBUG("[ScatterAutoSelector][%s] start, topoInfo levelNum[%u]", __func__, topoInfo.levelNum);
90 :
91 0 : if (topoInfo.levelNum > 1) {
92 0 : if (topoInfo.level0Shape == Level0Shape::MESH_1D) {
93 0 : if (topoInfo.netLayerDetails.localNetInsSizeOfLayer[0] == 1) {
94 0 : primQueueGenName = "InsScatterNHR";
95 : } else {
96 0 : primQueueGenName = "InsScatterParallelMesh1DNHR";
97 : }
98 0 : } else if (topoInfo.level0Shape == Level0Shape::MESH_2D) {
99 0 : HCCL_WARNING("[Algo][ScatterAutoSelector] level0Shape[%d] is not supported yet for aicpu levelNum > 1.",
100 : topoInfo.level0Shape);
101 0 : return SelectorStatus::NOT_MATCH;
102 0 : } else if (topoInfo.level0Shape == Level0Shape::CLOS) {
103 0 : HCCL_WARNING("[Algo][ScatterAutoSelector] level0Shape[%d] is not supported yet for aicpu levelNum > 1.",
104 : topoInfo.level0Shape);
105 0 : return SelectorStatus::NOT_MATCH;
106 : } else {
107 0 : return SelectorStatus::NOT_MATCH;
108 : }
109 : } else {
110 0 : if (topoInfo.level0Shape == Level0Shape::MESH_1D) {
111 0 : primQueueGenName = "InsScatterMesh1D";
112 0 : } else if (topoInfo.level0Shape == Level0Shape::MESH_2D) {
113 0 : primQueueGenName = "InsScatterMesh2D";
114 0 : } else if (topoInfo.level0Shape == Level0Shape::MESH_1D_CLOS) {
115 0 : if (IsLayerAllConnetedWithTopo(topoInfo, 0, TopoType::MESH_1D)) {
116 : // MESH_1D 即可链接所有卡， 使用 MESH_1D 算法
117 0 : primQueueGenName = "InsScatterMesh1D";
118 : } else {
119 0 : if (topoInfo.level0PcieMix) {
120 : // 预留PCIE mix入口，如果要更新算法可以直接改
121 0 : primQueueGenName = "InsScatterParallelMesh1DNHRPcie";
122 : }
123 0 : primQueueGenName = "InsScatterMesh1D";
124 : }
125 0 : } else if (topoInfo.level0Shape == Level0Shape::CLOS) {
126 0 : if (topoInfo.level0PcieMix) {
127 0 : primQueueGenName = "InsScatterNHR";
128 : } else {
129 0 : primQueueGenName = "InsScatterMesh1D";
130 : }
131 : } else {
132 0 : HCCL_WARNING("[ScatterAutoSelector] topo not match");
133 0 : return SelectorStatus::NOT_MATCH;
134 : }
135 : }
136 0 : HCCL_INFO("[Algo][ScatterAutoSelector][%s] Algo match [%s]", __func__, primQueueGenName.c_str());
137 0 : return SelectorStatus::MATCH;
138 : }
139 :
140 0 : SelectorStatus ScatterAutoSelector::SelectAivAlgo(const TopoInfo &topoInfo, const CollAlgOperator &op,
141 : const std::map<OpType, std::vector<HcclAlgoType>> &configAlgMap, std::string &primQueueGenName) const
142 : {
143 : (void)op;
144 : (void)configAlgMap;
145 0 : HCCL_DEBUG("[ScatterAutoSelector][%s] start, topoInfo levelNum[%u]", __func__, topoInfo.levelNum);
146 :
147 : // aiv 直接走打平 mesh
148 0 : primQueueGenName = "AivScatterMesh1D";
149 :
150 0 : HCCL_INFO("[Algo][ScatterAutoSelector][%s] Algo match [%s]", __func__, primQueueGenName.c_str());
151 0 : return SelectorStatus::MATCH;
152 : }
153 :
154 : REGISTER_SELECTOR_BY_OPTYPE(OpType::SCATTER, 18, ScatterAutoSelector);
155 : } // namespace Hccl
|
