`host_to_device_memcpy_sm` low than `device_to_host_memcpy_sm`
Nobody has claimed this yet.
- Dominant language
- C++
- Stars
- 778
- Forks
- 93
- PR merge metrics
- No merged PRs in 30d
Description
Hi!
When we tested host_to-vice_cemcpy_sm and Device_to-host_cemcpy_sm separately on the H100 cluster, we obtained two completely different values
Running host_to_device_memcpy_sm.
memcpy SM CPU(row) -> GPU(column) bandwidth (GB/s)
0 1 2 3 4 5 6 7
0 35.19 35.25 35.30 35.03 35.25 35.32 35.39 35.06
Running device_to_host_memcpy_sm.
memcpy SM CPU(row) <- GPU(column) bandwidth (GB/s)
0 1 2 3 4 5 6 7
0 52.77 52.77 52.77 52.78 52.76 52.77 52.78 52.77
Actually, they should be close values.
What could be causing this?
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 reproducing the separately reported host_to_device_memcpy_sm and device_to_host_memcpy_sm measurements on the H100 cluster, then inspect how each benchmark is configured and executed. Done means identifying the reason for the bandwidth difference and documenting or correcting the behavior, with updated measurements to verify the result.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- performance
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100