pytorch/torch/csrc/jit/frontend/source_ref.h
Han Qi b34b192d6b Reland "Make debug_pkl smaller by only emitting unique traces." (#73368)
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
2022-04-18 22:34:21 +00:00

48 lines
1.3 KiB
C++

#pragma once
#include <functional>
#include <memory>
#include <ATen/core/ivalue.h>
#include <c10/macros/Export.h>
#include <torch/csrc/jit/frontend/source_range.h>
namespace torch {
namespace jit {
/**
* SourceRef does two things:
* 1. Owns a Source object.
* 2. Serves as lookup key to the owned Source in associative containers, for
* runtime data aggregation.
* We don't want to use std::shared_ptr<Source> directly because we want to
* support heteogeneous lookup, and also shared_ptr is an implementation detail
* which should be encapsulated.
*/
class TORCH_API SourceRef : public CustomClassHolder {
public:
explicit SourceRef(std::shared_ptr<Source> source_view)
: source_view_(std::move(source_view)) {}
bool operator==(const SourceRef& other) const {
return source_view_ == other.source_view_;
}
bool operator<(const Source& other) const {
return source_view_.get() < &other;
}
friend bool operator<(const Source& other, const SourceRef& self) {
return &other < self.source_view_.get();
}
bool operator<(const SourceRef& other) const {
return *this < *other.source_view_.get();
}
const Source* operator->() const {
return source_view_.get();
}
private:
std::shared_ptr<Source> source_view_;
};
} // namespace jit
} // namespace torch