support ChainRulesCore inplaceability #2091
Merged
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
|
Thanks! Shouldn't this be part of a package extension though? |
Ah yes, that makes sense. I've changed this to be loaded via extensions (I simply adapted the code from SpecialFunctions to be consistent). |
piever
commented
Sep 26, 2023
| @@ -67,4 +70,5 @@ UnsafeAtomicsLLVM = "0.1" | |||
| julia = "1.8" | |||
|
|
|||
| [extras] | |||
| ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4" | |||
There was a problem hiding this comment.
I'm not sure if this is necessary, but I saw it was done for SpecialFunctions, so I added it here for consistency.
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Outdated suggestions cannot be applied.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.
ChainRulesCore has an opt-in trait
is_inplaceable_destinationto denote that an array supports in-place broadcasting (in particulara .+= b). This PR allows the ChainRules machinery to add derivatives in-place toCuArrays and avoid unnecessary allocations.I only added the method for
CuArrays because AFAIU wrapper types are handled by ChainRulesCore directly. Also, I didn't add the method for CUDA sparse arrays as it seemed to me that they do not support in-place broadcasting.This requires one extra dependency (ChainRulesCore), but it is very lightweight and it was already a transitive dependency via AbstractFFTs.