NVIDIA-NeMo / NVIDIA-NeMo/Automodel

CUDA Graphs support

Open
#551 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement
Dominant language
Python
Stars
963
Forks
318
Avg merge
3d 20h
Merged PRs (30d)
143

Description

Is your feature request related to a problem? Please describe.
Support for CUDA Graphs for

  • single GPU
  • multi GPU / multi Node

Describe the solution you'd like
Add support for capturing and replaying the graph. We can target static sequence inputs (optionally with packing).

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.

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

The issue names no files, tests, or entry points. First map the existing training execution and distributed single-GPU, multi-GPU, and multi-node paths, then determine how static sequence inputs and optional packing would be captured and replayed. Done means CUDA Graphs support is defined and working across the requested execution modes.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
Domain
distributed-systems, machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.