You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

single_op_model.h 3.1 kB

5 years ago
5 years ago
5 years ago
4 years ago
5 years ago
5 years ago
5 years ago
4 years ago
5 years ago
5 years ago
5 years ago
5 years ago
5 years ago
1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798
  1. /**
  2. * Copyright 2019-2020 Huawei Technologies Co., Ltd
  3. *
  4. * Licensed under the Apache License, Version 2.0 (the "License");
  5. * you may not use this file except in compliance with the License.
  6. * You may obtain a copy of the License at
  7. *
  8. * http://www.apache.org/licenses/LICENSE-2.0
  9. *
  10. * Unless required by applicable law or agreed to in writing, software
  11. * distributed under the License is distributed on an "AS IS" BASIS,
  12. * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13. * See the License for the specific language governing permissions and
  14. * limitations under the License.
  15. */
  16. #ifndef GE_SINGLE_OP_SINGLE_OP_MODEL_H_
  17. #define GE_SINGLE_OP_SINGLE_OP_MODEL_H_
  18. #include <cstdint>
  19. #include <map>
  20. #include <memory>
  21. #include <string>
  22. #include <vector>
  23. #include "common/helper/model_helper.h"
  24. #include "graph/load/new_model_manager/davinci_model_parser.h"
  25. #include "single_op/single_op.h"
  26. #include "single_op/stream_resource.h"
  27. namespace ge {
  28. struct SingleOpModelParam {
  29. uint64_t base_addr = 0;
  30. uint64_t memory_size = 0;
  31. uint64_t weight_addr = 0;
  32. uint64_t weight_size = 0;
  33. uint64_t zero_copy_mem_size = 0;
  34. uint8_t *mem_base = nullptr;
  35. uint8_t *weight_base = nullptr;
  36. std::map<uintptr_t, int> addr_mapping_;
  37. int64_t core_type = 0;
  38. };
  39. class SingleOpModel {
  40. public:
  41. SingleOpModel(const std::string &model_name,
  42. const void *model_data,
  43. uint32_t model_size);
  44. ~SingleOpModel() = default;
  45. Status Init();
  46. Status BuildOp(StreamResource &resource, SingleOp &single_op);
  47. Status BuildDynamicOp(DynamicSingleOp &single_op);
  48. private:
  49. Status InitModel();
  50. Status LoadAllNodes();
  51. Status ParseInputsAndOutputs();
  52. Status SetInputsAndOutputs(SingleOp &single_op);
  53. Status InitModelMem(StreamResource &resource);
  54. Status ParseInputNode(const OpDescPtr &op_desc);
  55. void ParseOutputNode(const OpDescPtr &op_desc);
  56. Status BuildTaskList(SingleOp &single_op);
  57. Status BuildTaskListForDynamicOp(DynamicSingleOp &dynamic_single_op);
  58. Status BuildKernelTask(const domi::KernelDef &kernel_def, TbeOpTask **task);
  59. Status BuildKernelExTask(const domi::KernelExDef &kernel_def, AiCpuTask **task,
  60. bool dynamic_flag, bool& depend_compute_flag, uint64_t kernel_id);
  61. Status BuildCpuKernelTask(const domi::KernelDef &kernel_def, OpTask **task, uint64_t kernel_id);
  62. Status BuildModelTaskKernel(const domi::TaskDef &task_def, DynamicSingleOp &single_op);
  63. static void ParseOpModelParams(ModelHelper &model_helper, SingleOpModelParam &param);
  64. void ParseArgTable(TbeOpTask *task, SingleOp &op);
  65. std::string model_name_;
  66. const void *ori_model_data_;
  67. uint32_t ori_model_size_;
  68. ModelHelper model_helper_;
  69. map<uint32_t, NodePtr> op_list_;
  70. SingleOpModelParam model_params_;
  71. std::vector<ptrdiff_t> input_offset_list_;
  72. std::vector<size_t> input_sizes_;
  73. std::vector<ptrdiff_t> output_offset_list_;
  74. std::vector<size_t> output_sizes_;
  75. std::vector<OpDescPtr> data_ops_;
  76. OpDescPtr netoutput_op_;
  77. bool has_weight_ = false;
  78. };
  79. } // namespace ge
  80. #endif // GE_SINGLE_OP_SINGLE_OP_MODEL_H_

图引擎模块(GE)是MindSpore的一个子模块,其代码由C++实现,位于前端模块ME和底层硬件之间,起到承接作用。图引擎模块以ME下发的图作为输入,然后进行一系列的深度图优化操作,最后输出一张可以在底层硬件上高效运行的图。GE针对昇腾AI处理器的硬件结构特点,做了特定的优化工作,以此来充分发挥出昇腾AI处理器的强大算力。在进行模型训练/推理时,GE会被自动调用而用户并不感知。GE主要由GE API和GE Core两部分组成,详细的架构图如下所示