Deleted tensor_core_tiled_layout.cu

[howardzhang-cv]

Stack from ghstack (oldest at bottom):

• -> Deleted tensor_core_tiled_layout.cu #3722

Summary: Deleted tensor_core_tiled_layout.cu
Deleted op registrations in ops.py
Deleted corresponding tests in test_ops.py
Deleted references in READMEs and setup.py

Test Plan:

Reviewers:

Subscribers:

Tasks:

Tags:

Differential Revision: D91500687

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/3722

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit 13a1468 with merge base ef944d2 ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[howardzhang-cv]

@howardzhang-cv has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[howardzhang-cv]

@howardzhang-cv has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[howardzhang-cv]

@howardzhang-cv has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[jerryzh168]

for context, we are deleting this for (1) better ABI (2) tensor core tiled layout is not as important as before as we have new hardwares, also dequantize op is not used in new stack: https://github.com/pytorch/ao/blob/main/torchao/quantization/quantize_/workflows/int4/int4_tile_packed_to_4d_tensor.py

[jerryzh168]

lg, please see comment inline

[howardzhang-cv]

@howardzhang-cv has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[howardzhang-cv]

@howardzhang-cv has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[andrewor14]

Looks great to me too! Thanks, merging this