"GPU Device list" is null object (regardless of GPU count) when json flag is used.
Nobody has claimed this yet.
- Dominant language
- C++
- Stars
- 778
- Forks
- 93
- PR merge metrics
- No merged PRs in 30d
Description
In non-multinode config, worldSize is set to "deviceCount" before cuDeviceGetCount(&deviceCount) is even called.
JsonOutput::recordDevices is given "worldSize" as a parameter, which is still 0, but the function implementation refers to this parameter as deviceCount.
The for loop exits immediately without populating the GPU device list in the json object.
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 tracing the non-multinode setup into JsonOutput::recordDevices and compare the worldSize argument with the value returned by cuDeviceGetCount. Reproduce the issue with the JSON flag and verify that the resulting GPU device list is populated for both zero and nonzero device-count paths.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- cli
- Issue type
- Bug
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 68/100