[Codegen] Use affine.delinearize_index in workgroup distribution #19839
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
In the interests of getting rid of needless substractions from affine composition, get rid of one of the last remaining manual calls to floorDiv() The other ProcInfo generators have been converted to delinearize_index by earlier PRs - this one finishes the job. This should not impact the behavior of generated programs.
hanhanW
reviewed
Feb 3, 2025
Co-authored-by: Han-Chung Wang <[email protected]>
|
@hanhanW Yeah, I'm switching us over to |
| // CHECK: %[[D20:.*]] = flow.dispatch.tensor.load %[[D5]], offsets = {{\[}}%[[ARG0]], %[[ARG1]], %[[ARG2]], %[[ARG3]]], sizes = {{\[}}%[[D9]], %[[D12]], %[[D15]], %[[D18]]], strides = [1, 1, 1, 1] : !flow.dispatch.tensor<readonly:tensor<?x?x?x?xf32>>{%[[D0]], %[[D1]], %[[D2]], %[[D3]]} -> tensor<?x?x?x?xf32> | ||
| // CHECK: %[[D21:.*]] = tensor.empty(%[[D9]], %[[D12]], %[[D15]], %[[D18]]) : tensor<?x?x?x?xf32> | ||
| // CHECK: %[[D22:.*]] = linalg.generic {indexing_maps = [#map9, #map9, #map9], iterator_types = ["parallel", "parallel", "parallel", "parallel"]} ins(%[[D19]], %[[D20]] : tensor<?x?x?x?xf32>, tensor<?x?x?x?xf32>) outs(%[[D21]] : tensor<?x?x?x?xf32>) attrs = {lowering_config = #config} { | ||
| // CHECK-DAG: %[[D7:.*]] = affine.apply #map(){{\[}}%[[D1]]] |
There was a problem hiding this comment.
Is there a reason you dont want to use linearize as well. These all seem like linearizations?
There was a problem hiding this comment.
Trying to keep the PR minimal but also the fact that a lot of that's coming out of the linalg machinery or otherwise isn't as easy to get to
Or at least that's my memory of why I didn't touch anything.
MaheshRavishankar
approved these changes
Feb 3, 2025
ita9naiwa
pushed a commit
to ita9naiwa/iree
that referenced
this pull request
Feb 4, 2025
…e-org#19839) In the interests of getting rid of needless substractions from affine composition, get rid of one of the last remaining manual calls to floorDiv() The other ProcInfo generators have been converted to delinearize_index by earlier PRs - this one finishes the job. This should not impact the behavior of generated programs. --------- Co-authored-by: Han-Chung Wang <[email protected]> Signed-off-by: Hyunsung Lee <[email protected]>
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.
In the interests of getting rid of needless substractions from affine composition, get rid of one of the last remaining manual calls to floorDiv()
The other ProcInfo generators have been converted to delinearize_index by earlier PRs - this one finishes the job.
This should not impact the behavior of generated programs.