Increase performances of the omp-for model - #3
Open
nicolas-bouton wants to merge 5 commits into
Open
Conversation
The guided OpenMP schedule kind doesn't provide the best performances. By running different tests, it is exhibited that the dynamic schedule kind achieves the best performances even with the default chunk size. However, more performance can be obtained by tweaking the chunk size, so this patch allows the user to decide at runtime which OpenMP schedule kind to use (along with its own chunk size). * omp-for/traversal.hh: Here. Signed-off-by: Nicolas Bouton <nicolas.bouton@sipearl.com>
Other models don’t measure how long deps arrays take to be created. * omp-for/traversal.hh: Here. Signed-off-by: Nicolas Bouton <nicolas.bouton@sipearl.com>
The timer was not restarted after calling the upwards function, thus the dtt timer was computed using the time spent in the upwards + deps + compute functions instead of deps + compute functions. For readability the deps and compute code has been moved into the omp_dtt function. * omp-for/traversal.hh: Here. Signed-off-by: Nicolas Bouton <nicolas.bouton@sipearl.com>
As with other model implementations. * omp-for/traversal.hh: Here. Signed-off-by: Nicolas Bouton <nicolas.bouton@sipearl.com>
* omp-for/traversal.hh: Here. Signed-off-by: Nicolas Bouton <nicolas.bouton@sipearl.com>
nicolas-bouton
force-pushed
the
nb/increase-omp-for-performance
branch
from
August 12, 2026 08:22
470e400 to
43d6514
Compare
Author
|
The commits were amended to provide additional context for each change. |
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The
guidedOpenMP schedule kind doesn't provide the best performances.By running different tests, it is exhibited that the
dynamicschedule kind achieves the best performances even with the default chunk size.However, more performance can be obtained by tweaking the chunk size, so this patch allows the user to decide at runtime which OpenMP schedule kind to use (along with its own chunk size).
Also, the timer was not restarted after calling the upwards function, thus the dtt timer was computed using the time spent in the upwards + deps + compute functions instead of deps + compute functions. For readability the deps and compute code has been moved into the omp_dtt function.