-
Notifications
You must be signed in to change notification settings - Fork 198
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
tweak the arena mr to reduce fragmentation #845
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice job. I think the byte formatting can be turned into a useful utility for the rest of RMM.
Co-authored-by: Mark Harris <[email protected]>
Also noticed copyrights need to be updated to include 2021. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Just need to update copyrights.
@rongou are you still working on this? |
Got busy with some other things. Will try to get back to this next week. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This looks good, thank you for adding the defragment
case on complete failure to allocate. @jlowe fyi.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good. Couple small comments.
std::size_t total_upstream{}; | ||
for (auto const& b : upstream_blocks_) { | ||
total_upstream += b.size(); | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Use accumulate
same as above.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Done.
@codereport please take another look. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
🔥
@gpucibot merge |
We're seeing out-of-memory errors in some cases when there are plenty of free space available, probably due to memory fragmentation. This PR attempts to address the issue:
With these changes I can run those TPC-DS queries successfully with very high concurrency that previously would run out of memory. Need to do some more benchmarking to verify the behavior.
@abellina