mlcommons / mlcommons/chakra

NCCL:Broadcast collectives are missing from the converted trace but present in the trace_link

Open
#161 5 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
196
Forks
84
PR merge metrics
No merged PRs in 30d

Description

Describe the Bug

After running a ResNet50 or TinyLlama2 workload on 4 ranks I see that in the Kineto trace at least one nccl:broadcast collective is observed. In the trace_link file the same collective is observed, but in the converted trace the collective is no longer present. Is this a normal behavior or is it an issue on the Chakra Converter side?

I looked in the converter implementation, but I did not observe any pointers that this should be done - dismiss broadcast collectives. Is there something I missed?

Steps to Reproduce

Using the Chakra version from 6 Sept, after the merge of commit #140.

Expected Behavior

See the nccl:broadcast collective in the converted trace.

Screenshots

This is the trace_link file, the broadcast collective is present.
Screenshot 2024-10-16 at 14 15 15
This is the converted trace, in json format, no broadcast collective can be found - search result is at the bottom of the picture.
Screenshot 2024-10-16 at 14 17 39

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

Reproduce the issue with a ResNet50 or TinyLlama2 workload on 4 ranks, then compare the Kineto trace, trace_link file, and converted JSON trace. The work is complete when the converted trace preserves the nccl:broadcast collective observed in the source traces.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
devtools
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
42/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.