How to get the communication details in SimCCL
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 1.2k
- Forks
- 184
- Avg merge
- 13h 4m
- Merged PRs (30d)
- 1
Description
In the paper S 3.4, there is a paragraph mentioning:
Collective Communication Transformation: SimCCL intercepts collective communication calls, reconstructing
the operations to trace lower-level communications. It
skips actual data transfers and captures inter-GPU communication events, including data size, sender and receiver ranks, and routes, to simulate RDMA-layer behavior
I wonder how to use SimCCL to get the inter-GPU communication events. Currently MultiDevOneP.cc does not seem to provide such information.
Contributor guide
No contributing guide indexed for this repository
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
Start by reading the communication-related code in MultiDevOneP.cc alongside the SimCCL description in paper section S 3.4. Determine whether the inter-GPU event details are exposed elsewhere or whether the usage is undocumented; done means providing a clear way to obtain data size, sender and receiver ranks, and routes, or documenting the missing capability.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- distributed-systems
- Issue type
- Documentation
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100