Add edge_interior_ratio argument to TilePartitioner #44
Conversation
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
…ng, added one test case with just one interface dimension
…nd fixed current unit tests.
…erm/pace into feature/shrinked_edge_partitioning
Contributor
|
Can one of the admins verify this patch? |
Collaborator
|
ok to test |
Contributor
|
launch jenkins |
2 similar comments
Contributor
|
launch jenkins |
Contributor
|
launch jenkins |
mcgibbon
reviewed
Dec 3, 2021
Collaborator
|
launch jenkins |
mcgibbon
reviewed
Dec 7, 2021
mcgibbon
reviewed
Dec 7, 2021
mcgibbon
reviewed
Dec 9, 2021
mcgibbon
reviewed
Dec 10, 2021
pace-util/HISTORY.md
Outdated
| - Renamed DummyComm to LocalComm, and added support for message tags. The DummyComm symbol is still in place for backwards compatibility, but points to LocalComm | ||
| - added error in CubedSphereCommunicator init if given a communicator with a size not equal to the total ranks of the given partitioner | ||
| - `subtile_extent` method of Partitioner classes now takes in a required `rank` argument | ||
| - TilePartitioner has a new `edge_interior_ratio` argument which defaults to 1.0, and lets the user specify the relative 1-dimensional extent of edge and corner compute domains compared to interior compute domains. In all cases, the closest valid value will be used, which enables some previously invalid configurations (e.g. C128 on a 3 by 3 layout will use the closest valid edge_interior_ratio to 1.0) |
Contributor
There was a problem hiding this comment.
Suggested change
| - TilePartitioner has a new `edge_interior_ratio` argument which defaults to 1.0, and lets the user specify the relative 1-dimensional extent of edge and corner compute domains compared to interior compute domains. In all cases, the closest valid value will be used, which enables some previously invalid configurations (e.g. C128 on a 3 by 3 layout will use the closest valid edge_interior_ratio to 1.0) | |
| - TilePartitioner has a new `edge_interior_ratio` argument which defaults to 1.0, and lets the user specify the relative 1-dimensional extent of the boundaries (edge and corner) compute domains compared to interior compute domains. In all cases, the closest valid value will be used, which enables some previously invalid configurations (e.g. C128 on a 3 by 3 layout will use the closest valid edge_interior_ratio to 1.0) |
Trying to align the wording with the code (e,g, ratio between interior and boundary tile sizes. in function comment). Boundary is a bit of a loaded term here, so you might want to flip for a constructed "edge-corner" concept or something better.
Contributor
|
Do we want to have at least one test using a non-default ratio that actually exchange data? Not quite a unit test per se, but it sounds safer to have at least one of those |
Collaborator
|
Good catch, I hadn't noticed there wasn't one. I'll add at least one each of a true MPI and mock MPI case. |
…-one edge_interior_ratio
Collaborator
|
launch jenkins |
twicki
pushed a commit
to twicki/pace
that referenced
this pull request
Aug 22, 2025
* initial commit, need to adapt and run tests * revising scheme name * tests pass * update history * linting * changing typehints for physics schemes to enum instead of str * driver now works with physics config enum, tests pass * fixed tests * missed one
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.
tests/test_partitioner.pyupdated to reflect some more advanced assignments.subtile_extentmethod of Partitioner classes now takes in a requiredrankargumentedge_interior_ratioargument which defaults to 1.0, and lets the user specify the relative 1-dimensional extent of edge and corner compute domains compared to interior compute domains. In all cases, the closest valid value will be used, which enables some previously invalid configurations (e.g. C128 on a 3 by 3 layout will use the closest valid edge_interior_ratio to 1.0)TilePartitioner.subtile_slice()to do custom subdomain partitioning.This PR is replacing #33 with its feature branch moved from my fork to the
pacerepo.