[inductor] TLParse tensor metadata logging + test by skarjala · Pull Request #160132 · pytorch/pytorch

skarjala · 2025-08-07T20:07:06Z

Summary:

Add TLParse artifact logging per op with output tensor shape, stride, and dtype for cross-rank aggregation.

Testing:

Add test to verify structure and contents of tlparse artifiact

cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @chenyang78 @kadeng @muchulee8 @amjames @chauhang @aakhundov @coconutruben @Lucaskabela

[ghstack-poisoned]

pytorch-bot · 2025-08-07T20:07:09Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/160132

📄 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 the bot commands wiki or our office hours

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

✅ You can merge normally! (1 Unrelated Failure)

As of commit 10e7567 with merge base 01bcf9a ():

FLAKY - The following job failed but was likely due to flakiness present on trunk:

pull / linux-jammy-py3.9-clang12 / test (crossref, 2, 2, lf.linux.2xlarge) (gh) (similar failure)
cpp/test_jit 1/1 failed!

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

ghstack-source-id: 86972c4 Pull-Request: #160132

[ghstack-poisoned]

pytorchmergebot · 2025-08-11T13:58:58Z

Starting merge as part of PR stack under #160260

pytorchmergebot · 2025-08-11T14:23:49Z

Starting merge as part of PR stack under #160260

[ghstack-poisoned]

ghstack-source-id: 1f9321b Pull-Request: #160132

xmfan · 2025-08-12T01:51:31Z

torch/_inductor/debug.py

+        trace_structured(
+            "artifact",
+            metadata_fn=lambda: {
+                "name": "inductor_tlparse_tensor_meta",


this should be combined with the existing runtime estimates artifact

xmfan · 2025-08-12T01:52:24Z

test/dynamo/test_structured_trace.py

+                    for out in op.get("outputs", []):
+                        self.assertIn("shape", out)
+                        self.assertIn("stride", out)
+                        self.assertIn("dtype", out)


please write this test using assertExpectedInline so that we can see the artifact's outputs from the test file

xmfan · 2025-08-12T01:53:08Z

test/dynamo/test_structured_trace.py

+        def fn(x):
+            y = x @ x
+            return y + 1


this is insufficient test coverage. at the very minimum, we should consider 1 case of each where shapes/stride/dtype are different between input and outputs

[ghstack-poisoned]

ghstack-source-id: 1f2c252 Pull-Request: #160132 fix pr feedback

[ghstack-poisoned]

ghstack-source-id: cad2dba Pull-Request: #160132 fix pr feedback update

xmfan · 2025-08-12T20:56:07Z

test/dynamo/test_structured_trace.py

+          "shape": [
+            2
+          ],
+          "stride": [
+            1
+          ]


we should include a test with dynamic shapes to test out the to_size_hints codepath

[ghstack-poisoned]

xmfan · 2025-08-13T16:23:55Z

test/dynamo/test_structured_trace.py

+            self.assertParses()
+
+    @requires_tlparse
+    @torch._dynamo.config.patch(dynamic_shapes=True)


this config doesnt do anything

xmfan · 2025-08-13T16:24:28Z

test/dynamo/test_structured_trace.py

+                w = z.to(torch.float16)
+                return w
+
+            compiled = torch.compile(f, backend="inductor", fullgraph=True)


either compile with dynamic=True or use mark_dynamic like in the starter tasks

[ghstack-poisoned]

xmfan · 2025-08-14T18:26:37Z

test/dynamo/test_structured_trace.py

+                simplified_ops = []
+                for op in ops:
+                    outs = [
+                        {
+                            "shape": out.get("shape", []),
+                            "stride": out.get("stride", []),
+                            "dtype": out.get("dtype", None),
+                        }
+                        for out in op.get("outputs", [])
+                    ]
+                    if outs:
+                        simplified_ops.append(
+                            {
+                                "type": op.get("type", ""),
+                                "outputs": outs,
+                            }
+                        )
+
+                simplified = (
+                    {"ops": simplified_ops[-1:]} if simplified_ops else {"ops": []}
+                )
+                actual = json.dumps(simplified, indent=2, sort_keys=True)


you could just self.assertExpectedInline(ops, ...)

xmfan · 2025-08-14T18:27:30Z

test/dynamo/test_structured_trace.py

+                self.assertExpectedInline(
+                    actual,
+                    r"""{
+  "ops": [


we should really add tests with at least multiple ops

ghstack-source-id: 11030a1 Pull-Request: #160132 fix pr feedback update add dynamic test fix dynamic test add tests w/ more ops fix cuda update change to cuda and triton fix cuda and triton

[ghstack-poisoned]

skarjala · 2025-08-16T16:35:14Z

@pytorchbot merge

pytorchmergebot · 2025-08-16T16:36:56Z

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

clee2000 · 2025-08-16T23:45:20Z

@pytorchbot revert -m "broke lint GH job link HUD commit link. landrace with another PR that changed some had_cuda related things" -c landrace

pytorchmergebot · 2025-08-16T23:46:50Z

@pytorchbot successfully started a revert job. Check the current status here.
Questions? Feedback? Please reach out to the PyTorch DevX Team

pytorchmergebot · 2025-08-16T23:47:06Z

@skarjala your PR has been successfully reverted.

This reverts commit 2603e40. Reverted #160132 on behalf of https://github.com/clee2000 due to broke lint [GH job link](https://github.com/pytorch/pytorch/actions/runs/17010600949/job/48226137423) [HUD commit link](https://hud.pytorch.org/pytorch/pytorch/commit/2603e40be5fa4a66301e6654e34a82a67f2e4913). landrace with another PR that changed some had_cuda related things ([comment](#160132 (comment)))

[ghstack-poisoned]

skarjala · 2025-08-17T04:09:30Z

@pytorchbot merge -i

pytorchmergebot · 2025-08-17T04:11:19Z

Merge started

Your change will be merged while ignoring the following 1 checks: pull / linux-jammy-py3.9-clang12 / test (crossref, 2, 2, lf.linux.2xlarge)

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

Summary: - Add TLParse artifact logging per op with output tensor shape, stride, and dtype for cross-rank aggregation. Testing: - Add test to verify structure and contents of tlparse artifiact Pull Request resolved: pytorch#160132 Approved by: https://github.com/xmfan ghstack dependencies: pytorch#160260

…0132)" This reverts commit 2603e40. Reverted pytorch#160132 on behalf of https://github.com/clee2000 due to broke lint [GH job link](https://github.com/pytorch/pytorch/actions/runs/17010600949/job/48226137423) [HUD commit link](https://hud.pytorch.org/pytorch/pytorch/commit/2603e40be5fa4a66301e6654e34a82a67f2e4913). landrace with another PR that changed some had_cuda related things ([comment](pytorch#160132 (comment)))

Summary: - Add TLParse artifact logging per op with output tensor shape, stride, and dtype for cross-rank aggregation. Testing: - Add test to verify structure and contents of tlparse artifiact Pull Request resolved: pytorch#160132 Approved by: https://github.com/xmfan

Summary: - Add TLParse artifact logging per op with output tensor shape, stride, and dtype for cross-rank aggregation. Testing: - Add test to verify structure and contents of tlparse artifiact Pull Request resolved: pytorch#160132 Approved by: https://github.com/xmfan ghstack dependencies: pytorch#160260

…0132)" This reverts commit 2603e40. Reverted pytorch#160132 on behalf of https://github.com/clee2000 due to broke lint [GH job link](https://github.com/pytorch/pytorch/actions/runs/17010600949/job/48226137423) [HUD commit link](https://hud.pytorch.org/pytorch/pytorch/commit/2603e40be5fa4a66301e6654e34a82a67f2e4913). landrace with another PR that changed some had_cuda related things ([comment](pytorch#160132 (comment)))

Summary: - Add TLParse artifact logging per op with output tensor shape, stride, and dtype for cross-rank aggregation. Testing: - Add test to verify structure and contents of tlparse artifiact Pull Request resolved: pytorch#160132 Approved by: https://github.com/xmfan

Update

ba939b3

[ghstack-poisoned]

pytorch-bot bot added ciflow/inductor module: dynamo module: inductor labels Aug 7, 2025

skarjala added a commit that referenced this pull request Aug 7, 2025

logging for tensor metadata

8522c42

ghstack-source-id: 86972c4 Pull-Request: #160132

skarjala changed the title ~~logging for tensor metadata~~ [inductor] TLParse tensor metadata logging + test Aug 7, 2025

skarjala requested review from xmfan and yushangdi August 7, 2025 20:10

Update

0cb110e

[ghstack-poisoned]

skarjala mentioned this pull request Aug 10, 2025

Fix collective schedule logging and runtime tests #160260

Closed

Update

169c051

[ghstack-poisoned]

skarjala added a commit that referenced this pull request Aug 11, 2025

logging for tensor metadata

3689d7c

ghstack-source-id: 1f9321b Pull-Request: #160132

xmfan reviewed Aug 12, 2025

View reviewed changes

Update

6084a92

[ghstack-poisoned]

skarjala added a commit that referenced this pull request Aug 12, 2025

logging for tensor metadata

26d56d0

ghstack-source-id: 1f2c252 Pull-Request: #160132 fix pr feedback

Update

a05d9ad

[ghstack-poisoned]

skarjala added a commit that referenced this pull request Aug 12, 2025

logging for tensor metadata

1e32466

ghstack-source-id: cad2dba Pull-Request: #160132 fix pr feedback update

skarjala mentioned this pull request Aug 12, 2025

[inductor] structured-log graph execution order + test #160448

Closed

xmfan reviewed Aug 12, 2025

View reviewed changes

Update

95c573e

[ghstack-poisoned]

xmfan reviewed Aug 13, 2025

View reviewed changes

Update

a99857c

[ghstack-poisoned]

xmfan reviewed Aug 14, 2025

View reviewed changes

xmfan approved these changes Aug 14, 2025

View reviewed changes

xmfan reviewed Aug 14, 2025

View reviewed changes

Update

ba0d17f

[ghstack-poisoned]

pytorchmergebot added the merging label Aug 16, 2025

pytorchmergebot closed this in 2603e40 Aug 16, 2025

pytorchmergebot added Merged and removed merging labels Aug 16, 2025

pytorchmergebot added Reverted ci-no-td Do not run TD on this PR labels Aug 16, 2025

pytorchmergebot reopened this Aug 16, 2025

Update

10e7567

[ghstack-poisoned]

pytorchmergebot added the merging label Aug 17, 2025

pytorchmergebot closed this in c699668 Aug 17, 2025

pytorchmergebot removed the merging label Aug 17, 2025

github-actions bot deleted the gh/skarjala/17/head branch September 17, 2025 02:07

Conversation

skarjala commented Aug 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pytorch-bot bot commented Aug 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/160132

✅ You can merge normally! (1 Unrelated Failure)

Uh oh!

pytorchmergebot commented Aug 11, 2025

Uh oh!

pytorchmergebot commented Aug 11, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

skarjala commented Aug 16, 2025

Uh oh!

pytorchmergebot commented Aug 16, 2025

Merge started

Uh oh!

clee2000 commented Aug 16, 2025

Uh oh!

pytorchmergebot commented Aug 16, 2025

Uh oh!

pytorchmergebot commented Aug 16, 2025

Uh oh!

skarjala commented Aug 17, 2025

Uh oh!

pytorchmergebot commented Aug 17, 2025

Merge started

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

skarjala commented Aug 7, 2025 •

edited

Loading

pytorch-bot bot commented Aug 7, 2025 •

edited

Loading