Patch/linear euler gpu resident #73
Draft
+1,446
−1,213
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
In recent multi-GPU runs, we found that MPI ranks were all piling work on the default device (device 0). This change sets the device ID for each MPI rank.
This is needed so that the built in source method doesn't run on GPU platforms. Since the source term for this model is zero currently, this just adds memcpy overhead that is unnecessary. This patch significantly improves model throughput on GPU platforms
Before allocating any GPU data, the domain decomposition must be initialized first when using multiple MPI ranks. This is due to the fact that the hipSetDevice call can potentially change the default device used for each rank. This patch fixes the broken CI run on this branch
Codecov ReportAttention: Patch coverage is
📢 Thoughts on this report? Let us know! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR improves the performance of the GPU accelerated implementation of the Linear Euler 2D application. By defining an empty
sourcemethodtype bound procedure in the GPU backend implementation, thesourcemethodis completely bypassed, which eliminates unnecessary CPU code execution and host-device memcpy's.