Interactive performance profiling and debugging tool for PyTorch neural networks.
Project description
DeepView.Profile is a tool to profile and debug the training performance of PyTorch neural networks.
Installation
DeepView.Profile works with GPU-based neural networks that are implemented in PyTorch.
To run DeepView.Profile, you need:
- A system equipped with an NVIDIA GPU
- Python 3.7+
- PyTorch 1.1.0+ with CUDA
- NOTE: We assume you have the correct version of PyTorch installed for their GPU. Default PyTorch installation on Linux distros might not have CUDA support. If you see error similar to below, your PyTorch version is incompatible with your version of CUDA. You can download the appropriate version from the PyTorch site
The current PyTorch install supports CUDA capabilities sm_37 sm_50 sm_60 sm_70. If you want to use the NVIDIA GeForce RTX 3060 Ti GPU with PyTorch, please check the instructions at https://pytorch.org/get-started/locally/
- NOTE: We assume you have the correct version of PyTorch installed for their GPU. Default PyTorch installation on Linux distros might not have CUDA support. If you see error similar to below, your PyTorch version is incompatible with your version of CUDA. You can download the appropriate version from the PyTorch site
Installation from PyPi
Installing with Pip
pip install deepview-profile
Installation from source
git clone https://github.com/CentML/DeepView.Profile
cd DeepView.Profile
poetry install
poetry run deepview --help
Usage example
To use DeepView.Profile in your project, you need to first write an entry point file, which is a regular Python file that describes how your model is created and trained. See the Entry Point for more information.
Once your entry point file is ready, there are two ways to profile interactive profiling and standalone profiling.
Interactive Profiling
Interactive profiling is done with VSCode with the DeepView.Explore plugin. Install the plugin in VSCode and run the profiling session to interactively profile your models.
python3 -m deepview_profile interactive
Standalone Profiling
Standalone profiling is useful when you just want access to DeepView.Profile's profiling functionality. DeepView.Profile will save the profiling results (called a "report") into a SQLite database file that you can then query yourself. We describe the database schema for DeepView.Profile's run time and memory reports in the Run Time Report Format and Memory Report Format pages respectively.
To have DeepView.Profile perform run time profiling, you use the deepview time
subcommand. In addition to the entry point file, you also need to specify the
file where you want DeepView.Profile to save the run time profiling report using the
--output
or -o
flag.
python3 -m deepview_profile time entry_point.py --output my_output_file.sqlite
Launching memory profiling is almost the same as launching run time profiling.
You just need to use deepview memory
instead of deepview time
.
python3 -m deepview_profile memory entry_point.py --output my_output_file.sqlite
Development Environment Setup
From the project root, do
poetry install
Release History
See Releases
Meta
DeepView.Profile began as a research project at the University of Toronto in collaboration with Geofrey Yu, Tovi Grossman and Gennady Pekhimenko.
The accompanying research paper appears in the proceedings of UIST'20. If you are interested, you can read a preprint of the paper here.
If you use DeepView.Profile in your research, please consider citing our paper:
@inproceedings{skyline-yu20,
title = {{Skyline: Interactive In-Editor Computational Performance Profiling
for Deep Neural Network Training}},
author = {Yu, Geoffrey X. and Grossman, Tovi and Pekhimenko, Gennady},
booktitle = {{Proceedings of the 33rd ACM Symposium on User Interface
Software and Technology (UIST'20)}},
year = {2020},
}
It is distributed under Apache 2.0 license. See LICENSE and NOTICE for more information.
Contributing
Check out CONTRIBUTING.md for more information on how to help with DeepView.Profile.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for deepview_profile-0.12.3-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 17eacc14af1b6e97f68399a979ffc1cd3829d1538f01f7892f4f88f9ff422e01 |
|
MD5 | 2281c72dcdff424848bfb28be9d5b31a |
|
BLAKE2b-256 | ea969eebfdcbb6beb09fdea757434585b7cbe7f1b70a3b1ab24cd46b3d5658cb |