Arm backend: Add passes to handle int64 const and int64 output ops #13803
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
- Add ConvertInt64ConstOpsToInt32Pass to convert constant-producing ops
that output int64 to instead output int32, when values are within
int32 bounds.
Supported Ops:
`torch.full`, `torch.arange`, `torch.eye`, `torch.linspace`,
`torch.tensor`
- Add ConvertInt64OutputOpsToInt32Pass to
1. convert or remove unnecessary casts to int64
2. insert an int64->int32 cast after the argmax ndoes
that produce int64 outputs
Signed-off-by: Yufeng Shi <yufeng.shi@arm.com>
Change-Id: I04e5fa9a7170c5b5dc785ae8619189545de0ec2c
Co-authored-by: Erik Lundell <erik.lundell@arm.com>
YufengShi-dudu
added
partner: arm
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/13803
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ❌ 4 New Failures, 4 Unrelated FailuresAs of commit ee352a4 with merge base 389918b ( NEW FAILURES - The following jobs have failed:
BROKEN TRUNK - The following jobs failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
bot
added
the
CLA Signed
|
Test fails are unrelated |
zingo
approved these changes
Sep 1, 2025
|
@YufengShi-dudu we've got a few test failures for this PR when it's tested internally with Buck.
Example log: |
|
Thanks for the issue and example log. We find that your internal Have you merged the int16 patches internally? We are also working on the int16. I think we will add some xfails to these tests as a turnaround before we improve data type checks during partitioning. cc: @Ninja91 @per |
|
@YufengShi-dudu I can check, which PR contains the int16 patches? |
|
@YufengShi-dudu This is the PR: #13658. It's merged and looks like waiting to be merged to main. I have added xfails on those tests as well. cc: @digantdesai |
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.
Add ConvertInt64ConstOpsToInt32Pass to convert constant-producing ops that output int64 to instead output int32, when values are within int32 bounds. Supported Ops:
torch.full,torch.arange,torch.eye,torch.linspace,torch.tensorAdd ConvertInt64OutputOpsToInt32Pass to
Change-Id: I04e5fa9a7170c5b5dc785ae8619189545de0ec2c
cc @digantdesai @freddan80 @per @zingo @oscarandersson8218