This PR runs an automatic check as part of dynamo_wrapped to make sure that all unimplemented_v2() callsites are mapped to the JSON file. It also fixes the issue of the CI not able to expand the hints, which was the root cause of the previous workflow failure. If not, the dev gets a message giving them instructions on how to update the JSON file. I also updated a dynamic gb_type to static and updated its test_error_message to include the GBID link for the graph break (before the link would not be produced).
Testing:
I ran the file with the argument to ensure all cases were covered, and also tested the test in CI.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/156237
Approved by: https://github.com/williamwen42
This PR includes the GBID weblink whenever a user encounters a graph break. I also had to include the JSON file in setup.py, so it can be part of the files that are packaged in during CI. It also fixes the issue of the hardcoded error messages stripping away one of the '/' in 'https'.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/156033
Approved by: https://github.com/williamwen42
This PR includes the GBID weblink whenever a user encounters a graph break. I also had to include the JSON file in setup.py, so it can be part of the files that are packaged in during CI. It also fixes the issue of the hardcoded error messages stripping away one of the '/' in 'https'.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/156033
Approved by: https://github.com/williamwen42
This word appears often in class descriptions and is not consistently spelled. Update comments and some function names to use the correct spelling consistently. Facilitates searching the codebase.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/155944
Approved by: https://github.com/Skylion007
This PR adds the workflow of whenever a dev makes a PR that contains files under torch/_dynamo, we check for any unimplemented_v2() callsites and if any of them have been modified in some sort of way, the workflow fails and lists them exactly which callsites and let's them know what the command lines are to update the registry.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/155610
Approved by: https://github.com/williamwen42
There is now functionality for the developer to add a --additional-info arg to the add and update dev terminal command to include any additional info the dev might want to remark about the graph break.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/155526
Approved by: https://github.com/williamwen42
The user can now use the terminal to update the registry whenever they update an existing gb_type's properties. Additionally, if the user changes the gb_type description itself, they can update the registry as well.
Terminal command template for updating existing gb_type: python [path to gb_id_mapping.py] update "existing_gb_type" [path to file where user added callsite]
Terminal command template for updating existing gb_type name (can also be used if the user changed the other properties as well including the gb_type name): python [path to gb_id_mapping.py] update "existing_gb_type" [path to file where user added callsite] --new_gb_type "new_name_for_existing_gb_type"
Pull Request resolved: https://github.com/pytorch/pytorch/pull/154985
Approved by: https://github.com/williamwen42
This allows for each device type to check current devices for Triton compatibility and ensure their Triton backend is present.
This PR replaces the `has_triton()` global method which was previously used for this task, and moves the initial check for each Inductor backend on to their associated `BaseScheduler` subclass. This means that other backends, such as Halide, can also implement their own availability checks.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139171
Approved by: https://github.com/jansel
Add triton support for ROCm builds of PyTorch.
* Enables inductor and dynamo when rocm is detected
* Adds support for pytorch-triton-mlir backend
* Adds check_rocm support for verify_dynamo.py
Pull Request resolved: https://github.com/pytorch/pytorch/pull/94660
Approved by: https://github.com/malfet
Add dynamo smoke tests to CI, which checks for python/torch/cuda versions and runs simple dynamo examples on a few backends, including inductor. Smoke tests will run on dynamo and inductor shards.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/89302
Approved by: https://github.com/malfet
