Implement a way to optionally start subclustering after all ranks have stopped swapping #486
Assignees
Labels
enhancement
New feature or request
Comments
ppebay
added a commit
that referenced
this issue
Dec 22, 2023
ppebay
added a commit
that referenced
this issue
Dec 22, 2023
ppebay
added a commit
that referenced
this issue
Dec 22, 2023
cwschilly
added a commit
that referenced
this issue
Jan 2, 2024
cwschilly
added a commit
that referenced
this issue
Jan 3, 2024
cwschilly
added a commit
that referenced
this issue
Jan 4, 2024
ppebay
added a commit
that referenced
this issue
Jan 10, 2024
…ks have stopped swapping (#488) * #486: fully separated swapping from subclustering and made other improvements * #486: fixed whitespace cleanup * #486: fixed a bug related to average load and removed intermediate variable * #486: completed optional separation of swapping and subclustering governed by an optional parameter * #486: fix subclustering test to reflect changes * #486: add unit test for new separate_subclustering param * #486: add missing config file --------- Co-authored-by: Caleb Schilly <[email protected]>
github-actions bot
pushed a commit
that referenced
this issue
Jan 10, 2024
…ks have stopped swapping (#488) * #486: fully separated swapping from subclustering and made other improvements * #486: fixed whitespace cleanup * #486: fixed a bug related to average load and removed intermediate variable * #486: completed optional separation of swapping and subclustering governed by an optional parameter * #486: fix subclustering test to reflect changes * #486: add unit test for new separate_subclustering param * #486: add missing config file --------- Co-authored-by: Caleb Schilly <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Currently the cluster swapping strategy (in non-deterministic mode) jumps into subclustering as soon as no full cluster swaps are possible on a given rank, irrespective of swapping/non-swapping on other ranks.
This has the effect of improving load balance locally but at the expense of earlier memory locality disruption than as compared to when subclustering is delayed until all ranks have exhausted full cluster swapping possibilities.
This is in contrast to the implementation of
temperedlbinvt, which has some global fence between cluster swaps and the subclustering stage.The goal of this issue is to allow for the latter (
vt-like) option in LBAF as well.@lifflander
The text was updated successfully, but these errors were encountered: