[LLVMGPU] Add im2col pipeline for convolution codegen #18086
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
Signed-off-by: Max Dawkins <max.dawkins@gmail.com>
Signed-off-by: Max Dawkins <max.dawkins@gmail.com>
Max191
requested review from
MaheshRavishankar,
qedawkins,
kuhar and
Groverkss
as code owners
This was referenced Aug 1, 2024
Signed-off-by: Max Dawkins <max.dawkins@gmail.com>
Max191
force-pushed
the
im2col-llvmgpu-pipeline
branch
from
1d75130
to
8e40a52
Compare
hanhanW
approved these changes
Aug 2, 2024
Comment on lines
+201
to
+202
| // CHECK: %[[LHS_RD:.+]] = vector.transfer_read %[[B0]]{{.*}} vector<8xf16> | ||
| // CHECK: vector.transfer_write %[[LHS_RD]] |
There was a problem hiding this comment.
This part is from im2col op, right?
There was a problem hiding this comment.
Yes, it should ultimately look the same as the normal matmul pipeline with different indexing on the LHS (im2col side). I don't bother checking the indexing here because the maps will likely get jumbled through canonicalizations and loop fusions throughout the pipeline.
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Outdated suggestions cannot be applied.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 PR adds the remaining needed passes for the IGEMM pipeline using the im2col op. It adds the
Conv2DToIm2colOppass with a flag--iree-codegen-llvmgpu-use-igemm, and it adds the im2col decomposition pass before vectorization passes.--iree-codegen-llvmgpu-use-igemmwill be false by default until the IGEMM pipeline is more robust and performant.