NVIDIA / NVIDIA/TransformerEngine

Reduce CPU overheads in te Sequential GroupedLinear Op

Open
#2,995 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
3.5k
Forks
831
Avg merge
3d 11h
Merged PRs (30d)
65

Description

Recently graph safe support has been added to te Sequential GroupedLinear Op https://github.com/NVIDIA/TransformerEngine/pull/2923.

But it suffers from CPU overheads. Nail the bottlenecks and fix them

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 by reading PR #2923 and tracing the graph-safe Sequential GroupedLinear Op path. Profile it to identify the CPU bottlenecks, then verify that the overhead is reduced without regressions in the operation’s existing behavior.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning, performance
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.