Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Resolve remaining LTC CI failures #1110

Merged
merged 13 commits into from
Jul 30, 2022

Conversation

henrytwo
Copy link
Member

@henrytwo henrytwo commented Jul 26, 2022

This PR addresses several issues currently faced by CI:

  1. Previously, LTC would refuse to build when TORCH_MLIR_USE_INSTALLED_PYTORCH=OFF due to LTC not checking this variable. The build scripts have been updated such that LTC uses the appropriate PyTorch build
  2. Blacklist any ops that are prefixed by new_ or end with _like in response to Make all of new_/_like factory functions composite explicit autograd pytorch/pytorch#82238
  3. Update LTC xfails to reflect new set of supported ops
  4. Prune unused Python C++ dependencies
  5. Add Shape inference function to var op, which is used by a recently added test
  6. Due to the aforementioned problems, and linkage issues with out-of-tree builds, this PR also modifies the CI to skip building LTC in the configs which currently do not succeed. This way, we can merge into main to ensure stability before investigating further.
    a. To avoid import issues from the LTC pybind no longer generating, this PR introduces a script to autogen a dummy Python file that allows for a successful import (however throws an error if it's actually used)

cc: @antoniojkim @ke1337

@henrytwo henrytwo self-assigned this Jul 26, 2022
@henrytwo henrytwo force-pushed the henrytu/ltc_use_installed_pt branch 5 times, most recently from dfe71ab to 7613155 Compare July 26, 2022 20:30
@henrytwo henrytwo changed the title Check value of TORCH_MLIR_USE_INSTALLED_PYTORCH during LTC build Resolve LTC CI failures Jul 26, 2022
@henrytwo henrytwo changed the title Resolve LTC CI failures Resolve remaining LTC CI failures Jul 26, 2022
@henrytwo henrytwo force-pushed the henrytu/ltc_use_installed_pt branch 3 times, most recently from e7af2b0 to b88264b Compare July 27, 2022 17:46
@henrytwo henrytwo force-pushed the henrytu/ltc_use_installed_pt branch 2 times, most recently from 431da1f to 18f9e35 Compare July 28, 2022 23:48
@henrytwo henrytwo marked this pull request as ready for review July 29, 2022 00:05
@henrytwo
Copy link
Member Author

henrytwo commented Jul 29, 2022

image
The failure from source build seems to be upstream related? I can see that it exists on main too

Based on the timing and errors, I think this is the root cause: pytorch/pytorch@2bfae07

We should still be good to merge though, since we saw source build succeed earlier prior to this PyTorch update

@vivekkhandelwal1
Copy link
Collaborator

image The failure from source build seems to be upstream related? I can see that it exists on main too

Based on the timing and errors, I think this is the root cause: pytorch/pytorch@2bfae07

We should still be good to merge though, since we saw source build succeed earlier prior to this PyTorch update

Yes, it's failing due to upstream PyTorch changes. #1121

@henrytwo henrytwo force-pushed the torch_mlir_ltc_backend branch 3 times, most recently from 54b3c5e to b149d13 Compare July 29, 2022 13:51
@henrytwo henrytwo force-pushed the henrytu/ltc_use_installed_pt branch from 5b44d86 to 6802f89 Compare July 29, 2022 13:53
@henrytwo
Copy link
Member Author

Looks like the latest failures are due to some changes on the side of Torch-MLIR -- fix coming shortly

@henrytwo henrytwo force-pushed the henrytu/ltc_use_installed_pt branch 2 times, most recently from b77b50d to 6cae84b Compare July 29, 2022 18:14
@powderluv
Copy link
Collaborator

Ok to land as long as CI is green

@henrytwo
Copy link
Member Author

Ok to land as long as CI is green

Thanks, can we get this PR stamped as approved then? I'll land this in tonight once CI passes

@henrytwo henrytwo mentioned this pull request Jul 29, 2022
Copy link
Contributor

@silvasean silvasean left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🚀

@henrytwo henrytwo merged commit 6d1db4c into torch_mlir_ltc_backend Jul 30, 2022
@henrytwo henrytwo deleted the henrytu/ltc_use_installed_pt branch July 30, 2022 12:16
henrytwo added a commit that referenced this pull request Jul 30, 2022
* Replace CHECK_EQ with TORCH_CHECK_EQ

* Check value of TORCH_MLIR_USE_INSTALLED_PYTORCH during LTC build

* Update LTC XFAIL with NewZerosModule ops

* Explicitly blacklist _like ops

* Automatically blacklist new_/_like ops

* Prune away unused Python dependencies from LTC

* Add flag to disable LTC

* Autogen dummy _REFERENCE_LAZY_BACKEND library when LTC is disabled

* Implement compute_shape_var

* Removed Var tests from XFAIL Set

* XFAIL tests using _local_scalar_dense or index.Tensor

* Add StdDim tests to XFAIL set

* Autogen aten::cat
qedawkins pushed a commit to nod-ai/torch-mlir that referenced this pull request Oct 3, 2022
Support RandomNormalLike ONNX operation.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants