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.

mem_allocator.h 1.6 kB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647
  1. /**
  2. * Copyright (c) Huawei Technologies Co., Ltd. 2022. All rights reserved.
  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 AIR_MEM_ALLOCATOR_H
  17. #define AIR_MEM_ALLOCATOR_H
  18. #include "exe_graph/runtime/tensor_data.h"
  19. #include "block.h"
  20. #include "exe_graph/runtime/allocator.h"
  21. namespace gert {
  22. namespace memory {
  23. struct MemAllocator {
  24. virtual Block *Malloc(size_t size) = 0;
  25. virtual ~MemAllocator() = default;
  26. };
  27. struct MemSynchronizer {
  28. MemSynchronizer() = default;
  29. virtual ~MemSynchronizer() = default;
  30. // Wait until the memory is actually freed after task completed
  31. virtual void Synchronize() const = 0;
  32. };
  33. } // namespace memory
  34. struct ExternalAllocators {
  35. public:
  36. memory::MemAllocator *GetAllocator(TensorPlacement placement, size_t usage);
  37. ge::Status SetAllocator(size_t placement, size_t usage, std::unique_ptr<memory::MemAllocator> allocator);
  38. private:
  39. std::unique_ptr<memory::MemAllocator> allocators[kTensorPlacementEnd][static_cast<size_t>(AllocatorUsage::kEnd)];
  40. };
  41. } // namespace gert
  42. #endif // AIR_CXX_MEM_ALLOCATOR_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