-
Notifications
You must be signed in to change notification settings - Fork 29
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
Trq/gpu multicopy #73
Conversation
Change-Id: I8544888b0e8f4105683cb29fc49d0ce9ecb97ad5
Finish gpu multicopy
DataManager::serializeRemoteChunk(): change name of parameter to avoid confusion with a local variable DataManager: Delete unused localParticles attribute Clean up/add documentation Some whitespace cleanup
N.B. My last commit was only tested on my laptop. It needs to be tested on an actual GPU. |
I'm realizing that all of the tests I've run so far have been with -DHAPI_TRACE enabled. Currently, this fails to build unless that flag is enabled. The CI doesn't catch this since it only does a non-cuda build. On a related note, I'm finding that 'testenergy' fails if HAPI_TRACE is disabled. I tried out a few different commits, and this issue extends at least back to the beginning of September before we started the project. I've only tried this on Frontera, so far, and it appears as if it might be specific to the SMP build (non-SMP works OK). |
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.
Missing #ifdef HAPI_TRACE
Also deleted a superfluous CmiWallTimer() call.
Make the local data transfer to GPU faster by using multiple threads to copy the data into the buffer.hh