WIP for scalar_mul_add #3454
Closed
WIP for scalar_mul_add #3454
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
pfultz2
reviewed
Sep 18, 2024
src/simplify_algebra.cpp
Outdated
| @@ -440,11 +440,44 @@ struct find_mul_add | |||
| auto x_ins = r.instructions["x"]; | |||
| assert(x_ins != b_ins); | |||
|
|
|||
| if(a_ins->get_shape().scalar()) | |||
There was a problem hiding this comment.
This should also check that it goes into convolution or gemm for the exception. It should also be done in the matcher.
aarushjain29
force-pushed
the
3432-improve-simplify_algebra-to-find-more-horizontal-fusion-opportunities
branch
from
e174012 to
7c2fdf5
Compare
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Outdated suggestions cannot be applied.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.
Draft PR
In the current example from SD Clip, there are 2 key kernels add_kernel and mul_add_kernel. After the computation of these kernels are completed, their results are fed into an mlir kernel which I think is a convolution.
I am implementing a scalar_mul_add in simplify_algebra. This will specifically handle where the multiplication operation is scalar and followed by a convolution. If these cases are met then directly perform a * (x+b) without expanding or rewriting it.
There needs to be an exception in mul_add. If the multiplication is scalar then it should not match and hence rewrite will not be done.