Conversation
Instead, check whether the script is under nsys via `NSYS_PROFILE_SESSION_ID`. Note that it's still possible to profile warmup iterations -- just don't specify `--capture-range cudaProfilerStart` in the `nsys` command.
mattteochen
left a comment
There was a problem hiding this comment.
Looks good, thank you.
|
I checked the CI errors and they are not related to this PR. Ready to merge! |
|
I wanted to read more about |
It's not official and that's a valid concern. I think it's OK because in the worst case we see an empty PS: it should be NSYS_PROFILING_SESSION_ID instead. However, that still isn't mentioned in any official documentation. |
kshitij12345
left a comment
There was a problem hiding this comment.
LGTM, thanks @wujingyue
t-vi
left a comment
There was a problem hiding this comment.
Thank you @wujingyue @mattteochen @kshitij12345
|
@KaelanDt I believe this is ready to merge. The CI failures look unrelated. |
Instead, check whether the script is under nsys via
NSYS_PROFILING_SESSION_ID. Note that it's still possible to profile warmup iterations -- just don't specify--capture-range cudaProfilerStartin thensyscommand.This partially rolls back #2661.