Skip to content
/ mpid2 Public

Utilities for the D2 CVMC group at the Max Planck Institute for Informatics

License

Notifications You must be signed in to change notification settings

mlosch/mpid2

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

30 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

mpid2

Utilities for the D2 CVMC group at the Max Planck Institute for Informatics

How-To

Install

python setup.py install or python setup.py develop

General note on dispatching on remote hosts

The underlying python scripts make use of a series of ssh calls. To circumvent repeated entering of the ssh password start an ssh agent:

exec ssh-agent bash ssh-add

Use pythond

  • bash scripts/gpujob_dispatching/install_pythond.sh
  • If ~/bin is in $PATH, then pythond can be called anywhere
  • To run a command on a free gpu/machine: pythond num_gpus gpu_mem cpu_mem command.py --args e.g. pythond 1 8 0 train.py, where memory is to be stated in GB.
  • Set the number of required cpu memory to 0, if unknown
  • To run a command on a free cpu machine set the number of required gpus to 0, e.g. pythond 0 0 256, to execute on cpu machines with at least 256GB of available RAM.

Use parallel job dispatching

Create a python script that sends a list of terminal commands to the next free hosts:

from mpid2 import dispatcher

commands = [
  'cd myworkingdirectory; bash preprocessing.sh && python train.py',
  'cd myworkingdirectory; bash preprocessing.sh && python train_2.py',
  ]
  
 hosts = ['host1', 'host2', 'host3', 'host4']
 
 dispatcher.dispatch(hosts, commands, required_gpus=2, required_gpu_mem=8, required_cpu_mem=0, log_target='file')

About

Utilities for the D2 CVMC group at the Max Planck Institute for Informatics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published