deepspeedai / deepspeedai/DeepSpeed
[REQUEST] Replace reduce in ZERO 1/2/3 with reduce_scatter
Open
Nobody has claimed this yet.
enhancement
- Dominant language
- Python
- Stars
- 43.1k
- Forks
- 5k
- Avg merge
- 4d 15h
- Merged PRs (30d)
- 112
Description
In ZeRO 1/2/3 with unchanged computational graph(ignore the change of shape), replacing all_reduce with reduce_scatter and carefully dividing the optimizer parameters can reduce the time of reduction operation.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
The issue names no files or tests. Start by locating the ZeRO 1/2/3 reduction paths and studying how all_reduce, reduce_scatter, and optimizer parameters interact; done means the replacement preserves the computational graph while reducing the reduction time across all three stages.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- distributed-systems, machine-learning, performance
- Issue type
- Refactor
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100