Support for UG kernels. #2579
Merged
Support for UG kernels. #2579
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
|
@LaurentMazare I just came across ug today and that led me to this pr in Candle. it would be great if you could share some details about ug. From ug readme we can see it's similar to tinygrad, so this will enable dynamic graph optimisations for candle as I know candle works in eager mode. |
|
It's pretty early days at the moment but the goal is indeed to provide a backend to candle that would allow for improved fusion. ug is somewhere between triton and tinygrad and to start with candle will use it more in a triton way to make it easier to write new kernels that work on all of cpu/metal/cuda. |
EricLBuehler
pushed a commit
to EricLBuehler/candle
that referenced
this pull request
Nov 26, 2024
* Support for UG kernels. * Add a dedicated test.
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Outdated suggestions cannot be applied.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.
This adds some preliminary support for kernels implemented using UG, this will allow for easy fusion of kernels. Only cuda is supported in this PR, the metal backend hasn't been activated yet.