-
Notifications
You must be signed in to change notification settings - Fork 1.8k
[fix][test] clear cuda cache before unittests automatically #5121
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[fix][test] clear cuda cache before unittests automatically #5121
Conversation
/bot run |
PR_Github #8436 [ run ] triggered by Bot |
/bot kill |
c676235
to
b031afa
Compare
/bot run |
PR_Github #8441 [ kill ] triggered by Bot |
PR_Github #8436 [ run ] completed with state |
PR_Github #8441 [ kill ] completed with state |
PR_Github #8445 [ run ] triggered by Bot |
PR_Github #8445 [ run ] completed with state |
/bot run --disable-fail-fast |
PR_Github #8501 [ run ] triggered by Bot |
b031afa
to
8102c47
Compare
/bot run |
PR_Github #8537 [ run ] triggered by Bot |
PR_Github #8537 [ run ] completed with state |
8102c47
to
647f992
Compare
/bot run |
PR_Github #9095 [ run ] triggered by Bot |
/bot kill |
PR_Github #9104 [ kill ] triggered by Bot |
PR_Github #9095 [ run ] completed with state |
PR_Github #9104 [ kill ] completed with state |
647f992
to
85e6939
Compare
/bot run |
PR_Github #9193 [ run ] triggered by Bot |
PR_Github #9193 [ run ] completed with state |
/bot run |
PR_Github #9202 [ run ] triggered by Bot |
PR_Github #9202 [ run ] completed with state |
5efca1d
to
1abca40
Compare
/bot run |
PR_Github #9244 [ run ] triggered by Bot |
PR_Github #9244 [ run ] completed with state |
1abca40
to
38eab7d
Compare
/bot run |
PR_Github #9250 [ run ] triggered by Bot |
PR_Github #9250 [ run ] completed with state |
/bot run |
PR_Github #9306 [ run ] triggered by Bot |
PR_Github #9306 [ run ] completed with state |
/bot run |
PR_Github #9410 [ run ] triggered by Bot |
Signed-off-by: Omer Ullman Argov <[email protected]>
ed2b4ab
to
09929bd
Compare
/bot run |
PR_Github #9411 [ run ] triggered by Bot |
PR_Github #9410 [ run ] completed with state |
PR_Github #9411 [ run ] completed with state |
Clear torch CUDA cache before unittests
We've encountered a case in which a test failed due to OOM errors, that were resolved by adding
torch.cuda.empty_cache
at the start of the test. This PR adds this to all unittests, so each one starts with an empty torch cuda cache and can make full use of the available device memory.