Skip to content

Clean up README.md and add requirements.py #209

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Clean up README.md and add requirements.py #209

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
28 changes: 12 additions & 16 deletions et_replay/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,29 +1,25 @@
# Execution Trace Replay (et_replay)
`et_replay` is a tool designed for replaying Chakra Execution Traces (ET) from machine learning models.
# License
Chakra replay is released under the Apache 2.0 license. Please see the LICENSE file for more information.

# Execution Trace Replay (chakra_replay)
`chakra_replay` is a tool designed for replaying Chakra Execution Traces (ET) from machine learning models.

## Installation
To install `et_replay`, use the following commands:
To install `chakra_replay`, use the following commands:

```bash
$ git clone --recurse-submodules [email protected]:facebookresearch/param.git
$ conda create -n et_replay python=3.10
$ conda activate et_replay
$ cd param
$ git clone --recurse-submodules [email protected]:pytorch-labs/chakra_replay.git
$ conda create -n chakra_replay python=3.10
$ conda activate chakra_replay
$ cd chakra_replay
$ pip3 install -r requirements.txt
$ cd et_replay
$ pip3 install .
```

## Running et_replay
To use et_replay, execution traces are required.
Start by collecting an execution trace using the command below. This command runs a benchmark with specific configurations and enables execution tracing.
```bash
$ python -m param_bench.train.compute.python.pytorch.run_benchmark -c train/compute/python/examples/pytorch/configs/simple_add.json --et
```

After collecting the trace, replay it with the following command. Set the warm-up iteration count to at least 1 to exclude tensor transfer time to GPUs.
Replay the trace with the following command. Set the warm-up iteration count to at least 1 to exclude tensor transfer time to GPUs.
```bash
$ python -m et_replay.tools.et_replay --input <trace_path> --warmup-iter 10 --iter 50 --compute --profile-replay
$ python -m tools.et_replay --input <trace_path> --warmup-iter 10 --iter 50 --compute --profile-replay
```

> Note: When analyzing performance values from et_replay, refer to the collected Kineto traces rather than the execution time reported by et_replay. Kineto traces are only collected when --profile-replay is provided.
4 changes: 4 additions & 0 deletions et_replay/requirements.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,4 @@
torch
future
numpy
pydot