mirror of
https://github.com/pytorch/pytorch.git
synced 2025-10-20 21:14:14 +08:00
b34b192d6b
Summary: ## Original commit message: Pull Request resolved: https://github.com/pytorch/pytorch/pull/73368 debug_pkl file inside of pytorch's .pt file consists of a list of SourceRanges. Each SourceRange points to a Source which is a stack track, filename, and start, end numbers. Those are emitted in debug_pkl file as strings. Since many SourceRange shares the same source, the string for trace can be deduped. The newer format saves a set of unique traces in a tuple, then each SourceRange will save the offset of it's trace w.r.t. position in that tuple. (i.e. manually applying dictionary compression). The above helps with smaller file size. On loading, if we copy each trace to Source as string the runtime memory would still blowup. To mitigate this, we use SourceView directly instead of source which will take the reference of string inside of Deserializer and make that into string_view. This is safe because Deserializer is hold by Unpickler by shared_ptr, and Unpickler is also hold by shared_ptr by another Source object. That Source object will be alive during the model construction. Test Plan: ## Original Test plan unit test Took original file (312271638_930.predictor.disagg.local); loaded with `torch.jit.load` save again with `torch.jit.save`. Unzip both, look at contents: ``` [qihan@devvm5585.vll0 ~]$ du archive -h 4.0K archive/xl_model_weights 3.7M archive/extra 8.0K archive/code/__torch__/caffe2/torch/fb/model_transform/splitting 8.0K archive/code/__torch__/caffe2/torch/fb/model_transform 8.0K archive/code/__torch__/caffe2/torch/fb 8.0K archive/code/__torch__/caffe2/torch 8.0K archive/code/__torch__/caffe2 20M archive/code/__torch__/torch/fx/graph_module 20M archive/code/__torch__/torch/fx 8.0K archive/code/__torch__/torch/classes 20M archive/code/__torch__/torch 20M archive/code/__torch__ 20M archive/code 2.7M archive/constants 35M archive [qihan@devvm5585.vll0 ~]$ du resaved -h 4.0K resaved/extra 8.0K resaved/code/__torch__/caffe2/torch/fb/model_transform/splitting 8.0K resaved/code/__torch__/caffe2/torch/fb/model_transform 8.0K resaved/code/__torch__/caffe2/torch/fb 8.0K resaved/code/__torch__/caffe2/torch 8.0K resaved/code/__torch__/caffe2 1.3M resaved/code/__torch__/torch/fx/graph_module 1.3M resaved/code/__torch__/torch/fx 8.0K resaved/code/__torch__/torch/classes 1.4M resaved/code/__torch__/torch 1.4M resaved/code/__torch__ 1.4M resaved/code 2.7M resaved/constants 13M resaved [qihan@devvm5585.vll0 ~]$ ``` ## Additional test: `buck test mode/dev-tsan //caffe2/benchmarks/static_runtime:static_runtime_cpptest -- --exact 'caffe2/benchmarks/static_runtime:static_runtime_cpptest - StaticRuntime.to'` passes test jest.fbios.startup_cold_start.local.simulator f333356873 - Differential Revision: D35196883 Pull Request resolved: https://github.com/pytorch/pytorch/pull/74869 Approved by: https://github.com/gmagogsfm
PyTorch JIT
This folder contains (most of) the C++ code for the PyTorch JIT, a language and compiler stack for executing PyTorch models portably and efficiently. To learn more about the JIT from a user perspective, please consult our reference documentation and tutorials.
A brief summary of the source tree:
- OVERVIEW.md: High-level technical overview of the JIT.
- frontend/: Taking PyTorch modules in Python and translating them into the JIT IR.
- ir/: Core IR abstractions.
- runtime/: Interpreter, graph execution, and JIT operators.
- codegen/: Generating efficient, hardware-specific code for JIT subgraphs.
- serialization/: Saving and loading modules.
- api/: Any user-facing C++ or Python interfaces.
- python/: Binding stuff into Python or accessing information from the Python environment.
- testing/: Utilities and helpers for testing.
- mobile/: Mobile-specific implementations of runtime components.
- passes/: IR-to-IR passes, generally for optimization and lowering.
- generated/: This folder is generated by the PyTorch build, and contains bindings for native PyTorch operators into the JIT.
Refer to each folder for more in-depth documentation.
Other relevant parts of the codebase not contained here:
- aten/src/ATen/core: contains JIT code re-used by other elements of the runtime system (eager, mobile, etc.)