Extend lookahead to cover host-buffer allocations for send / receive #313
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
We want to replace strided-host-buffer staging of MPI transfers with linearized copy_plan staging eventually, but in the meantime we must avoid frequent host-buffer reallocations (and with it, potential runaway scheduling complexity) around sends and receives.
|
Check-perf-impact results: (c42c71d6dbf9df683e01725118dd2b84) Relative execution time per category: (mean of relative medians)
|
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
Pull Request Test Coverage Report for Build 12051046056Details
💛 - Coveralls |
PeterTh
approved these changes
Nov 27, 2024
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Outdated suggestions cannot be applied.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.
Multi-node RSim surfaced the fact that we still generate frequent small allocations because we do not
anticipate()push and await-push commands, which do allocate (buffer) host memory.We want to replace strided-host-buffer staging of MPI transfers with linearized
copy_planstaging eventually, but for the meantime this PR avoids frequent host-buffer reallocations (and with it, potential runaway scheduling complexity) around sends and receives by accumulating all boxes they potentially touch duringanticipate()as well.