rccl icon indicating copy to clipboard operation
rccl copied to clipboard

Topo explorer failure to compile fix

Open alex-breslow-amd opened this issue 5 months ago • 0 comments

Details

Do not mention proprietary info or link to internal work items in this PR.

Work item: "Internal", or link to GitHub issue (if applicable). Internal

What were the changes?
Fix topo_explorer

Why were the changes made?
We need this tool to work.

How was the outcome achieved?
Fix problems albeit in a hacky way.

Additional Documentation:
It now compiles and runs. Yay. The main issues are an nvls call is not defined at link time. We could create a stub for that in topo explorer or hipify the file the nvls function is included in. I think neither of these is worth it to be honest, but I can make the change if reviewers see fit.

Another problem was that the file now needs something that only exists post ROCm 6.0. So we include the ROCm version. This means topo_expl won't compile pre-ROCm 6.0. I'm actually fine with that. This tool is for us (the RCCL developers).

Approval Checklist

Do not approve until these items are satisfied.

  • [ ] Verify the CHANGELOG has been updated, if
    • there are any NCCL API version changes,
    • any changes impact library users, and/or
    • any changes impact any other ROCm library.

alex-breslow-amd avatar Sep 05 '25 22:09 alex-breslow-amd