[PyTorch pin update] 951c21d6 - Inductor tests are starting to fail after pybind11 update to 2.13.6

[pbchekin]

PyTorch commit id: 951c21d6790334d57862e94a3f582ac724147a53
https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/10879914823/job/30185587514

_________ KernelTests.test_layout_constraint_needs_fixed_stride_order __________
Traceback (most recent call last):
  File "/runner/_work/intel-xpu-backend-for-triton/intel-xpu-backend-for-triton/pytorch/test/inductor/test_triton_kernels.py", line 1072, in test_layout_constraint_needs_fixed_stride_order
    eager_out = f(x)
  File "/runner/_work/intel-xpu-backend-for-triton/intel-xpu-backend-for-triton/pytorch/test/inductor/test_triton_kernels.py", line 1068, in f
    arange_out(x, y)
  File "/runner/_work/intel-xpu-backend-for-triton/intel-xpu-backend-for-triton/pytorch/test/inductor/test_triton_kernels.py", line 1059, in arange_out
    kernel[grid](x, out, n_elements, BLOCK_SIZE=4)
  File "/opt/hostedtoolcache/Python/3.9.20/x64/lib/python3.9/site-packages/triton/runtime/jit.py", line 330, in <lambda>
    return lambda *args, **kwargs: self.run(grid=grid, warmup=False, *args, **kwargs)
  File "/opt/hostedtoolcache/Python/3.9.20/x64/lib/python3.9/site-packages/triton/runtime/jit.py", line 657, in run
    kernel = self.compile(
  File "/opt/hostedtoolcache/Python/3.9.20/x64/lib/python3.9/site-packages/triton/compiler/compiler.py", line 315, in compile
    metadata_group[metadata_filename] = fn_cache_manager.put(json.dumps(metadata, default=vars), metadata_filename,
  File "/opt/hostedtoolcache/Python/3.9.20/x64/lib/python3.9/json/__init__.py", line 234, in dumps
    return cls(
  File "/opt/hostedtoolcache/Python/3.9.20/x64/lib/python3.9/json/encoder.py", line 199, in encode
    chunks = self.iterencode(o, _one_shot=True)
  File "/opt/hostedtoolcache/Python/3.9.20/x64/lib/python3.9/json/encoder.py", line 257, in iterencode
    return _iterencode(o, 0)
TypeError: vars() argument must have __dict__ attribute

[pbchekin]

CC @alexbaden

[anmyachev]

Reducing the number of potential problems:

• db393fb95e5b057ca49472828bb6dba2db4f859e (works)
• 426580a67db15ec17b2b861a09667bf59927e033 (failed)
• 8df01c82587f80b2800c6cfcbabe168d96cba731 (works)
• first failed commit: pytorch/pytorch@7f5abb4. " Update pybind11 to 2.13.6"
  • Details: the root cause of the error is new self._pybind11_conduit_v1_() method (new feature of pybind 2.13.6: https://pybind11.readthedocs.io/en/latest/changelog.html#version-2-13-6-september-13-2024) in GPUTarget class

[anmyachev]

For reference: pybind/pybind11#5377

[pbchekin]

Do you think we can temporary pin pybind11 to the last working version?

[anmyachev]

Do you think we can temporary pin pybind11 to the last working version?

This needs to be done on the PyTorch side by changing the version of pybind submodule, which can make a big difference from what they will use. So I'm not sure that's an option.

[vlad-penkin]

Blocked by:

• [Inductor] Fix Triton tests after updating pybind11 to 2.13.6 pytorch/pytorch#136280

[alexbaden]

pybind fix was merged last night/this morning, so this should also be resolved.

[vlad-penkin]

We need to remove patch in our ci and local pytorch wheel builds before closing this ticket.

[anmyachev]

Removed in #2351