Fix some typed storage is deprecated warnings.

[ezyang]

Stack from ghstack (oldest at bottom):

• -> Fix some typed storage is deprecated warnings. #89867

Signed-off-by: Edward Z. Yang ezyang@fb.com

cc @zou3519 @Chillee @samdow @soumith

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/89867

• 📄 Preview Python docs built from this PR
• 📄 Preview C++ docs built from this PR
• ❓ Need help or want to give feedback on the CI? Visit our office hours

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

❌ 1 Failures

As of commit 3958eb5:

The following jobs have failed:

• cuda11.6-py3.10-gcc7-sm86 / test (inductor_timm, 2, 2, linux.g5.4xlarge.nvidia.gpu)

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

[albanD]

SGTM
If you want to test this, you can do the same as this test:

pytorch/test/test_mps.py

Lines 7171 to 7211 in a029ec2

	def test_warn_on_not_implemented_with_fallback(self):
	_, _, _, op = self._get_not_implemented_op()
	script = f"""
	import os
	# MUST happen before pytorch's import
	os.environ["PYTORCH_ENABLE_MPS_FALLBACK"] = "1"
	import warnings
	with warnings.catch_warnings(record=True) as w:
	import torch
	if len(w) > 0:
	print(w)
	exit(1)
	# This should run just fine and raise warning about perf
	with warnings.catch_warnings(record=True) as w:
	{op}
	if len(w) != 1:
	print(w)
	exit(2)
	"""
	try:
	subprocess.check_output(
	[sys.executable, '-W', 'all', '-c', script],
	stderr=subprocess.STDOUT,
	# On Windows, opening the subprocess with the default CWD makes `import torch`
	# fail, so just set CWD to this script's directory
	cwd=os.path.dirname(os.path.realpath(__file__)),)
	except subprocess.CalledProcessError as e:
	if e.returncode == 1:
	self.assertTrue(False, "There was a warning when importing torch when PYTORCH_ENABLE_MPS_FALLBACK is set." +
	e.output.decode("utf-8"))
	elif e.returncode == 2:
	self.assertTrue(False, "There wasn't exactly one warning when running not implemented op with "
	f"PYTORCH_ENABLE_MPS_FALLBACK set. {e.output}")
	else:
	self.assertTrue(False, "Running a not implemented op failed even though PYTORCH_ENABLE_MPS_FALLBACK is set. " +
	e.output.decode("utf-8"))

[kurtamohler]

Just double checking that you remember that Tensor._storage() gives an UntypedStorage, whereas Tensor.storage() gave a TypedStorage. But if these changes aren't causing any failures, then that's probably okay

[ezyang]

yes, I want untyped here

[ezyang]

@pytorchbot merge

[pytorchmergebot]

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging

Check the merge workflow status
here

[pytorchmergebot]

Merge failed

Reason: 2 additional jobs have failed, first few of them are: inductor ,inductor / cuda11.6-py3.10-gcc7-sm86 / test (inductor_timm, 2, 2, linux.g5.4xlarge.nvidia.gpu)

Details for Dev Infra team

Raised by workflow job

[ezyang]

@pytorchbot merge -f "flaky ci"

[pytorchmergebot]

Merge started

Your change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging

Check the merge workflow status
here