NVIDIA-NeMo / NVIDIA-NeMo/Run

Support `ntasks` in `SlurmExecutor`

Open
#381 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement
Dominant language
Python
Stars
258
Forks
113
Avg merge
1d 3h
Merged PRs (30d)
5

Description

Is your feature request related to a problem? Please describe.

The current JobGroup relies on the first executor task to build arguments for sbatch.
It would be convenient to support the sbatch --ntasks argument so that the resource allocation calculation is a little easier than assigning --ntasks-per-node.

Describe the solution you'd like

Add ntasks to SlurmExecutor.

Describe alternatives you've considered

The alternative is to manually calculate potential topology of tasks per node based on the list of executors added to the experiment.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the SlurmExecutor and JobGroup implementations, then trace how executor settings become sbatch arguments. Add support for configuring ntasks and verify that the resulting submission includes the corresponding --ntasks resource request.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
infrastructure
Issue type
Feature
Difficulty
3/5
Estimated time
1-2 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
65/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.