MAINT,ENH: Refactor kernel argument handling into KernelArguments class#9705
Open
MAINT,ENH: Refactor kernel argument handling into KernelArguments class#9705
KernelArguments class#9705Conversation
This comment was marked as outdated.
This comment was marked as outdated.
1 similar comment
Member
Author
|
/test mini |
Member
Author
|
/test mini |
This reorganizes kernel argument handling to package them into a custom object. That has small advantages: 1. It allows passing them around slightly more conveniently. 2. It gives us a place to centralize at least _some_ functionality. 3. Having a mutable owning object makes it cleaner and easier to mutate a single owner (could be a list) rather than juggle many. This is already not bad, but some may actually go a bit too far in this commit.
(It would be good to move this though!)
seberg
force-pushed
the
args-refactor
branch
from
adc032b to
2e964ed
Compare
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This is a bit related to gh-9586 and the previous attempt on it in gh-2920. I.e. to make ElementWise kernels and ufuncs (and reductions) more similar one part of the old PR was also to introduce a new
Argsclass to encapsulate kernel arguments.This PR introduces a
KernelArgumentsand merges the broadcasting logic intofind_and_apply_shape().There is still quite a bit of duplication, but it helps to make the main
__call__paths easier to follow, I think. I opted for.get_in(0)getters for many things, because I would like to hide a bit how the arguments are stored.Should reduce overheads for simple ufuncs by around 5-10%, but for me it is more about helping with follow-ups on merging the ufunc/elementwise kernel code or just moving more things into
KernelArguments.Two concrete thoughts I have here is that
KernelArgumentscould already createCArrays, which could give us a pretty big boost when broadcasting orreduce_dims()happens (avoid creating morendarrays).Another thought I have is to speed up
ArgInfos a bit by merging them (the cache lookup takes quite a lot longer than it should).