[Codegen][CPU] Enable scalable transfer lowerings #18170
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
MacDue
force-pushed
the
enable_scalable_transfers
branch
3 times, most recently
from
09a58cd
to
b28f396
Compare
This enables a general scalable lowering for `transfer_write(transpose)` when ArmSME is _not_ available. The ArmSME dialect already had its own (more specific) lowerings for cases like this, which is why these lowerings are disabled when SME is available. Depends on: llvm/llvm-project#101353 Signed-off-by: Benjamin Maxwell <[email protected]>
Signed-off-by: Benjamin Maxwell <[email protected]>
MacDue
force-pushed
the
enable_scalable_transfers
branch
from
b28f396
to
645ab7d
Compare
hanhanW
reviewed
Aug 15, 2024
| // CHECK-LABEL: func.func @scalable_transpose_store | ||
| // CHECK-NOT: vector.transpose | ||
| // CHECK: vector.store {{.*}} : memref<?x?xf32>, vector<4xf32> | ||
| // CHECK-NOT: vector.transpose |
There was a problem hiding this comment.
Why do we need the second CHECK-NOT?
There was a problem hiding this comment.
CHECK-NOT only checks between two matches. The first checks between func.func @scalable_transpose_store and vector.store, the second checks from vector.store to the end of the function (IIRC).
There was a problem hiding this comment.
Yeah, I understand what it is happening. My point is that the check of vector.store already shows that the lowering happens. In this case, why do we need to check if there is a vector.transpose followed by it?
There was a problem hiding this comment.
I also want to check that the transpose (which is not directly supported) is eliminated.
hanhanW
approved these changes
Aug 15, 2024
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 enables a general scalable lowering for
transfer_write(transpose)when ArmSME is not available. The ArmSME dialect already had its own (more specific) lowerings for cases like this, which is why these lowerings are disabled when SME is available.Depends on: llvm/llvm-project#101353