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.

ops_kernel_builder.h 1.9 kB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556
  1. /**
  2. * Copyright 2020 Huawei Technologies Co., Ltd
  3. * Licensed under the Apache License, Version 2.0 (the "License");
  4. * you may not use this file except in compliance with the License.
  5. * You may obtain a copy of the License at
  6. * http://www.apache.org/licenses/LICENSE-2.0
  7. * Unless required by applicable law or agreed to in writing, software
  8. * distributed under the License is distributed on an "AS IS" BASIS,
  9. * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  10. * See the License for the specific language governing permissions and
  11. * limitations under the License.
  12. */
  13. #ifndef INC_COMMON_OPSKERNELUTILS_OPS_KERNEL_INFO_UTILS_H_
  14. #define INC_COMMON_OPSKERNELUTILS_OPS_KERNEL_INFO_UTILS_H_
  15. #include "external/ge/ge_api_error_codes.h"
  16. #include "cce/aicpu_engine_struct.h"
  17. #include "common/opskernel/ops_kernel_info_types.h"
  18. #include "graph/node.h"
  19. #include "proto/task.pb.h"
  20. namespace ge {
  21. class OpsKernelBuilder {
  22. public:
  23. OpsKernelBuilder() = default;
  24. virtual ~OpsKernelBuilder() = default;
  25. // initialize OpsKernelBuilder
  26. virtual Status Initialize(const std::map<std::string, std::string> &options) = 0;
  27. // finalize OpsKernelBuilder
  28. virtual Status Finalize() = 0;
  29. // memory allocation requirement
  30. virtual Status CalcOpRunningParam(Node &node) = 0;
  31. // generate task for op
  32. virtual Status GenerateTask(const Node &node, RunContext &context,
  33. std::vector<domi::TaskDef> &tasks) = 0;
  34. // only call aicpu interface to generate task struct
  35. virtual Status GenSingleOpRunTask(const NodePtr &node, STR_FWK_OP_KERNEL &task, string &task_info) {
  36. return FAILED;
  37. }
  38. // only call aicpu interface to generate task struct
  39. virtual Status GenMemCopyTask(uint64_t count, STR_FWK_OP_KERNEL &task, string &task_info) {
  40. return FAILED;
  41. }
  42. };
  43. } // namespace ge
  44. #endif // INC_COMMON_OPSKERNELUTILS_OPS_KERNEL_INFO_UTILS_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