open-mmlab / open-mmlab/mmengine
[Feature] disable distributed pytorch
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 1.5k
- Forks
- 464
- PR merge metrics
- No merged PRs in 30d
Description
What is the feature?
The default pytorch implementation on the nvidia orin does not include torch.distributed. So to use mmengine on the orin requires a special build of pytorch which is non-trivial. So, it would be awesome if there was a way to disable distributed so the code doesn't look for functions which aren't present: ReduceOp() for example.
Any other context?
No response
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
No files or tests are identified in the issue. Start by locating MMEngine code paths that call torch.distributed, then determine where distributed support could be disabled for PyTorch builds without those functions. Done means Orin users can run MMEngine without torch.distributed while supported distributed configurations continue to work.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- distributed-systems, machine-learning
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 30/100