Skip to content
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

[WIP] CUDA memory allocation #7506

Merged
merged 49 commits into from Apr 17, 2019

Conversation

Projects
None yet
2 participants
@alexanderst
Copy link
Contributor

commented Apr 10, 2019

This PR changes behaviour of nd4j-cuda backend to get rid of legacy "periodic gc" mechanics.

raver119 added some commits Apr 12, 2019

...
- smaller host cache
- purge before sleeping
Merge branch 'master' into asto_cuda_mm
# Conflicts:
#	nd4j/nd4j-backends/nd4j-backend-impls/nd4j-cuda/src/main/java/org/nd4j/nativeblas/Nd4jCuda.java
#	nd4j/nd4j-backends/nd4j-backend-impls/nd4j-native/src/main/java/org/nd4j/nativeblas/Nd4jCpu.java
- back to multiple deallocator threads per device
- purgeCache on oom regardless of free space
- ADSI always uses workspace by default
- periodicGc off by default

@raver119 raver119 merged commit 9e68fd8 into master Apr 17, 2019

1 check was pending

Codacy/PR Quality Review Hang in there, Codacy is reviewing your Pull request.
Details

@raver119 raver119 deleted the asto_cuda_mm branch Apr 17, 2019

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
You can’t perform that action at this time.