Adding ttir.repeat op in MLIR #1941
Open
+240
−24
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
sdjordjevicTT
requested review from
jnie-TT,
kmabeeTT,
AleksKnezevic,
pilkicTT,
svuckovicTT,
mtopalovicTT,
jserbedzijaTT,
azecevicTT,
nsmithtt,
mrakitaTT and
tapspatel
as code owners
sdjordjevicTT
force-pushed
the
sdjordjevic/adding_repeat_op
branch
from
c81dd9a to
3cfbbcc
Compare
sdjordjevicTT
force-pushed
the
sdjordjevic/adding_repeat_op
branch
from
3cfbbcc to
4026340
Compare
|
Please add an EmitC test as well :) |
|
FYI @ashokkumarkannan1, one this change lands on MLIR main, we should map it to the FFE and test it out :)) |
mtopalovicTT
approved these changes
Jan 23, 2025
| // RepeatOp | ||
| //===----------------------------------------------------------------------===// | ||
|
|
||
| // BroadcastOp verification |
Yep, I will add it! :) |
azecevicTT
approved these changes
Jan 23, 2025
| %input = ... : tensor<2x3xf32> | ||
|
|
||
| // Repeat each dimension twice | ||
| %repeated = "repeat"(%input) {repeat_dimensions = dense<[2, 2]> : tensor<2xi64>} : tensor<2x3xf32> -> tensor<4x6xf32> |
There was a problem hiding this comment.
Nit: ttir.repeat
Dense is written as: array<i32: 2, 2>
| @@ -803,6 +803,40 @@ def TTIR_ConcatOp : TTIR_DPSOp<"concat"> { | |||
| let hasVerifier = 1; | |||
| } | |||
|
|
|||
| def TTIR_RepeatOp : TTIR_DPSOp<"repeat"> { | |||
There was a problem hiding this comment.
As we have discussed offline, I think both solutions are fine (broadcast vs repeat), I just believe we should have some note that they are currently semantically the same, either in the comment or in the op description.
| ::mlir::RankedTensorType outputType = getOutput().getType(); | ||
| llvm::ArrayRef<int32_t> repeatDimensions = getRepeatDimensions(); | ||
|
|
||
| // Input tensor and repeate dimension argument must have same rank |
There was a problem hiding this comment.
Nit: dot at the end of the comment.
|
|
||
| // ----- | ||
|
|
||
| // Verify that the parsing fails if the input tensor and repeat_dimensions attribute doesn't have the same rank |
There was a problem hiding this comment.
Comment should reflect behaviour.
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.
Implementing TTIR repeat op. This PR lowers TTIR repeat op to TTNN repeat op. Added compiler and silicon tests for repeat op.
I used this opportunity to group tests for data movement ops under the same folder.
Closes #1916