oaoni / raytuneLSF

Python wrapper for starting distributed Ray clusters with LSF job submission

raytuneLSF

Run distributed ray applications with multinode LSF job submission.

Requirements

ray == 1.4.1

Getting started

Clone repo and cd into the project directory

$ conda create -n rayenv python=3.7
$ conda activate rayenv
$ pip install ray 
$ git clone https://github.com/oaoni/raytuneLSF.git
$ cd raytuneLSF

Run demo ray application:

Local

$ python ray_app.py -L

Distributed via LSF job submission

$ python run_job.py\
 -j RayJob\
 --nworkers 2\
 --cpusperworker 2\
 --mempercpu "4"\
 --walltime "0:20"