[Torch] Add support for aten.round.decimals op #4166
Conversation
This file contains hidden or 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
Contributor
mmanzoorTT
commented
Apr 28, 2025
•
mmanzoorTT
commented
- Added decomposition to aten.round
- Added test to projects/pt1/python/torch_mlir_e2e_test/test_suite/elementwise.py
vivekkhandelwal1
changed the title
vivekkhandelwal1
requested changes
Apr 30, 2025
There was a problem hiding this comment.
I have some comments:
1.) The PR title was incorrect. It said [TORCH] Add support for aten.reduce.decimals op while the op added is aten.round.decimals op. Also, it says that the support is added but only a folder is added.
2.) The shape and dtype has not been added. Please do that.
3.) Add the lowering for the op which lowers to Linalg.
4.) Add an e2e test for the op.
mmanzoorTT
force-pushed
the
tenstorrent/torch-round-decimals
branch
4 times, most recently
from
bc8dad7 to
48005db
Compare
|
@vivekkhandelwal1 Thanks a lot for the feedback. I have updated the PR as suggested. Please have a look |
vivekkhandelwal1
requested changes
May 16, 2025
Comment on lines
694
to
730
| if (auto round = dyn_cast<AtenRoundDecimalsOp>(op)) { | ||
| // AtenRoundDecimalsOp is decomposed, if decimals is non-zero, as follow. | ||
| // scale = 10 ** decimals | ||
| // return round(x * scale) / scale | ||
| if (!isa<mlir::FloatType>( | ||
| cast<ValueTensorType>(round.getType()).getDtype())) { | ||
| round.emitError("unimplemented: non-floating point dtype"); | ||
| return nullptr; | ||
| } | ||
| int64_t decimals; | ||
| if (!matchPattern(op->getOperand(1), m_TorchConstantInt(&decimals))) { | ||
| round.emitError("non-constant decimal point is not supported."); | ||
| return nullptr; | ||
| } | ||
|
|
||
| Value newOp = payloadArgs[0]; | ||
| Value scale; | ||
| if (decimals) { | ||
| auto elementType = | ||
| cast<RankedTensorType>( | ||
| converter->convertType(op->getOperand(0).getType())) | ||
| .getElementType(); | ||
|
|
||
| auto scalaVal = static_cast<float>(pow(10, decimals)); | ||
| scale = b.create<arith::ConstantOp>( | ||
| loc, FloatAttr::get(elementType, scalaVal)); | ||
| newOp = b.create<arith::MulFOp>(loc, newOp, scale); | ||
| } | ||
|
|
||
| newOp = b.create<math::RoundEvenOp>(loc, newOp); | ||
|
|
||
| if (decimals) { | ||
| newOp = b.create<arith::DivFOp>(loc, newOp, scale); | ||
| } | ||
|
|
||
| return newOp; | ||
| } |
There was a problem hiding this comment.
It's better to add it to DecomposeComplexOps.cpp.
There was a problem hiding this comment.
Thanks for the suggestion @vivekkhandelwal1 . updated.
mmanzoorTT
force-pushed
the
tenstorrent/torch-round-decimals
branch
2 times, most recently
from
1684186 to
f460ca2
Compare
* Added decomposition to aten.round * Added test to projects/pt1/python/torch_mlir_e2e_test/test_suite/elementwise.py
mmanzoorTT
force-pushed
the
tenstorrent/torch-round-decimals
branch
from
f460ca2 to
fbc9352
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.