coll/tuned: Extend the collective tuning file to be topology-aware #12321
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
wenduwan
reviewed
Feb 8, 2024
jiaxiyan
force-pushed
the
tuned_dynamic_file
branch
from
d710370
to
86e25c3
Compare
wenduwan
requested changes
Feb 9, 2024
TUNED collectives selection should account for communicator topology like HAN. The communicator size and message based algorithm selection logic is no longer sufficient to achieve optimal performance when HAN is used. The best algorithm differs between inter-node and intra-node for the same communicator size and message size based on the tuning results. This commit introduces topology dimension in both TUNED collective tuning file rule and the algorithm selection logic. The topological level can be intra-node, internode, or default(mixed). Specify @inter_node or @intra_node after the message size in the dynamic file rules. This is an optional feature so it will not break the old file format. See the file example in coll_tuned_dynamic_file.h Signed-off-by: Jessie Yang <jiaxiyan@amazon.com>
jiaxiyan
force-pushed
the
tuned_dynamic_file
branch
from
86e25c3
to
4c6b561
Compare
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Outdated suggestions cannot be applied.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.
TUNED collectives selection should account for communicator topology like HAN. The communicator size and message based algorithm selection logic is no longer sufficient to achieve optimal performance when HAN is used. The best algorithm differs between inter-node and intra-node for the same communicator size and message size based on the tuning results.
This commit introduces communicator topology dimension in both TUNED collective tuning file rule and the algorithm selection logic. The topological level can be single node, disjoint, or default(mixed).
Specify @single_node or @ disjoint after the message size in the dynamic file rules. This is an optional feature so it will not break the old file format. See the file example in coll_tuned_dynamic_file.h