https://github.com/shunk031/autocvd-go
A golang cli tool for setting `CUDA_VISIBLE_DEVICES` based on GPU utilization.
https://github.com/shunk031/autocvd-go
Last synced: 7 months ago
JSON representation
A golang cli tool for setting `CUDA_VISIBLE_DEVICES` based on GPU utilization.
- Host: GitHub
- URL: https://github.com/shunk031/autocvd-go
- Owner: shunk031
- License: mit
- Created: 2023-03-25T14:01:05.000Z (over 2 years ago)
- Default Branch: master
- Last Pushed: 2023-03-26T07:14:54.000Z (over 2 years ago)
- Last Synced: 2025-01-10T17:53:46.189Z (9 months ago)
- Language: Go
- Homepage:
- Size: 13.7 KB
- Stars: 2
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# autocvd-go
[](https://github.com/shunk031/autocvd-go/actions/workflows/ci.yaml)
[](https://github.com/shunk031/autocvd-go/actions/workflows/release.yaml)
A golang cli tool for setting `CUDA_VISIBLE_DEVICES` based on GPU utilization.
This project is heavily inspired by the python version of [`jonasricker/autocvd`](https://github.com/jonasricker/autocvd).
The aim of this tool, as well as [`jonasricker/autocvd`](https://github.com/jonasricker/autocvd), is as follows:
> On a system with multiple NVIDIA GPUs, autocvd eliminates the need for manually specifying the CUDA_VISIBLE_DEVICES environment variable. This comes in especially handy on systems with multiple users, like a shared GPU server. It is dependency-free and requires no code changes.
## Requirements
autocvd-go uses `nvidia-smi` to query GPU utilization. Make sure that it is installed and callable.
## Installation
Download the binary from [GitHub Releases](https://github.com/shunk031/autocvd-go/releases/latest) and drop it in your $PATH.
```shell
wget https://github.com/shunk031/autocvd-go/releases/latest/download/autocvd_linux_x86_64.tar.gz \
&& tar -xvzf autocvd_linux_x86_64.tar.gz autocvd \
&& rm autocvd_linux_x86_64.tar.gz
# Then, move `autocvd` binary in your $PATH
```
## Usage
To execute a command on a free GPU as well as [`jonasricker/autocvd`](https://github.com/jonasricker/autocvd), run the following:
```console
$ eval $(autocvd)
```
Possible use cases when training deep learning models that use GPUs include the following:
```console
$ eval $(autocvd) python train.py
```
## Examples
This example is identical to [`jonasricker/autocvd`](https://github.com/jonasricker/autocvd).
```shell
# run command on two free GPUs
$ eval $(autocvd -n 2)
# run command on least-used GPU (i.e., do not wait if no GPU is free)
$ eval $(autocvd -l)
# exclude certain GPUs
$ eval $(autocvd -x 0 2)
# if no free GPU is available immediately, wait for 60 seconds only
$ eval $(autocvd -t 60)
# export environment variables into the current shell
$ . <(autocvd -e) # alternatively: source <(autocvd -e)
```
## LICENSE
MIT
## Acknowledgment
- jonasricker/autocvd: Tool to automatically set CUDA_VISIBLE_DEVICES based on GPU utilization. Usable from command line and code. https://github.com/jonasricker/autocvd#examples