Skip to content
New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

Setting maximum roundable bin size to 4M #149

Merged
merged 1 commit into from
May 27, 2016

Conversation

drnikolaev
Copy link

CUB rounds allocation size up to next power of 2 to increase cache hit probability. This overhead becomes a problem when we make allocations of, let say, 2.5G. We do 4G instead which is a waste. This fix goes from 2^31 down to 2^22 (4MB).

@borisfom
Copy link

You realize it also decreases maximum cached block size to the same, right ?

@drnikolaev drnikolaev merged commit 2f0fbed into NVIDIA:caffe-0.15 May 27, 2016
@drnikolaev drnikolaev mentioned this pull request May 28, 2016
@drnikolaev drnikolaev deleted the caffe-0.15-fix branch June 7, 2016 06:48
# for free to join this conversation on GitHub. Already have an account? # to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants