Skip to main content

A queue service for quickly developing scripts that use all your GPUs efficiently

Project description

fastgpu

A queue service for quickly developing scripts that use all your GPUs efficiently

fastgpu provides a single command, fastgpu_poll, which polls a directory to check for scripts to run, and then runs them on the first available GPU. If no GPUs are available, it waits until one is. If more than one GPU is available, multiple scripts are run in parallel, one per GPU.

An API is also provided for polling programmatically, which is extensible for assigning other resources to processes besides GPUs.

Install

pip install fastgpu

How to use

--help provides command help:

$ fastgpu_poll --help
usage: fastgpu_poll [-h] [--path PATH] [--exit EXIT]

Poll `path` for scripts using `ResourcePoolGPU.poll_scripts`

optional arguments:
  -h, --help   show this help message and exit
  --path PATH  Path containing `to_run` directory
  --exit EXIT  Exit when `to_run` is empty

path defaults to the current directory. The path should contain a subdirectory to_run containing executable scripts you wish to run. It should not contain any other files, although it can contain subdirectories (which are ignored).

fastgpu_poll will run each script in to_run in sorted order. Each script will be assigned to one GPU. The id of the GPU selected will be passed as the first argument to the script; it is the responsibility of your script to use this GPU.

Once a script is selected to be run, it is moved into a directory called running. Once it's finished, it's moved into complete or fail as appropriate. stdout and stderr are captured to files with the same name as the script, plus stdout or stderr appended.

If exit is 1 (which is the default), then once all scripts are run, fastgpu_poll will exit. If it is 0 then fastgpu_poll will continue running until it is killed; it will keep polling for any new scripts that are added to to_run.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for fastgpu, version 0.0.2
Filename, size File type Python version Upload date Hashes
Filename, size fastgpu-0.0.2-py3-none-any.whl (9.5 kB) File type Wheel Python version py3 Upload date Hashes View hashes
Filename, size fastgpu-0.0.2.tar.gz (5.0 kB) File type Source Python version None Upload date Hashes View hashes

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page