Test shape analysis with opinfos #59814

eellison · 2021-06-10T19:27:28Z

Stack from ghstack:

Test shape analysis with opinfos #59814 Test shape analysis with opinfos
add ssupport for a few more opinfos in jit #59812 add ssupport for a few more opinfos in jit
Don't substitute in symbolic shapes to shape compute graph #59811 Don't substitute in symbolic shapes to shape compute graph
small cleanups #59810 small cleanups
Only optimize after change (redo) #59809 Only optimize after change (redo)

Using opinfos to test shape analysis. By default, we just check that we don't give incorrect answers, and then if assert_jit_shape_analysis is true, tests that we correctly propagates the full shape. and it found a couple bugs 😃

Differential Revision: D30200058

[ghstack-poisoned]

facebook-github-bot · 2021-06-10T19:27:34Z

🔗 Helpful links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/59814
📄 Preview docs built from this PR

💊 CI failures summary and remediations

As of commit 3bdbaf1 (more details on the Dr. CI page):

4/4 failures possibly* introduced in this PR
- 1/4 non-scanned failure(s)

🕵️ 2 new failures recognized by patterns

The following CI failures do not appear to be due to upstream breakages:

pytorch_xla_linux_bionic_py3_6_clang9_build (1/2)

Step: "(Optional) Merge target branch" (full log | diagnosis details | 🔁 rerun)

Automatic merge failed; fix conflicts and then commit the result.

CONFLICT (add/add): Merge conflict in torch/csrc/jit/python/init.cpp
Auto-merging torch/csrc/jit/python/init.cpp
CONFLICT (add/add): Merge conflict in torch/csrc/jit/passes/symbolic_shape_analysis.cpp
Auto-merging torch/csrc/jit/passes/symbolic_shape_analysis.cpp
CONFLICT (add/add): Merge conflict in torch/csrc/jit/passes/remove_mutation.cpp
Auto-merging torch/csrc/jit/passes/remove_mutation.cpp
CONFLICT (add/add): Merge conflict in torch/_C/__init__.pyi.in
Auto-merging torch/_C/__init__.pyi.in
CONFLICT (add/add): Merge conflict in test/test_ops.py
Auto-merging test/test_ops.py
Automatic merge failed; fix conflicts and then commit the result.


Exited with code exit status 1

pytorch_linux_xenial_py3_6_gcc5_4_build (2/2)

Step: "(Optional) Merge target branch" (full log | diagnosis details | 🔁 rerun)

Automatic merge failed; fix conflicts and then commit the result.

CONFLICT (add/add): Merge conflict in torch/csrc/jit/python/init.cpp
Auto-merging torch/csrc/jit/python/init.cpp
CONFLICT (add/add): Merge conflict in torch/csrc/jit/passes/symbolic_shape_analysis.cpp
Auto-merging torch/csrc/jit/passes/symbolic_shape_analysis.cpp
CONFLICT (add/add): Merge conflict in torch/csrc/jit/passes/remove_mutation.cpp
Auto-merging torch/csrc/jit/passes/remove_mutation.cpp
CONFLICT (add/add): Merge conflict in torch/_C/__init__.pyi.in
Auto-merging torch/_C/__init__.pyi.in
CONFLICT (add/add): Merge conflict in test/test_ops.py
Auto-merging test/test_ops.py
Automatic merge failed; fix conflicts and then commit the result.


Exited with code exit status 1

1 failure not recognized by patterns:

Job	Step	Action
^{pytorch_linux_bionic_cuda10_2_cudnn7_py3_9_gcc7_test2}	^{Set Up CI Environment After attach_workspace}	🔁 rerun

ci.pytorch.org: 1 failed

Failed: pr/pytorch-linux-bionic-rocm4.2-py3.6

This comment was automatically generated by Dr. CI (expand for details).

Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions to the (internal) Dr. CI Users group.

Click here to manually regenerate this comment.

ghstack-source-id: 106ae52 Pull Request resolved: #59814

Using opinfos to test shape analysis. By default, we just check that we don't give incorrect answers, and then if `assert_jit_shape_analysis` is true, tests that we correctly propagates the full shape. and it found a couple bugs 😃 [ghstack-poisoned]

test/test_ops.py

torch/testing/_internal/common_methods_invocations.py

Using opinfos to test shape analysis. By default, we just check that we don't give incorrect answers, and then if `assert_jit_shape_analysis` is true, tests that we correctly propagates the full shape. and it found a couple bugs 😃 [ghstack-poisoned]

mruberry · 2021-07-21T07:06:20Z

OpInfo-related changes look good in this PR but see comments in #59812 about a couple tweaks that should make the changes even simpler

Using opinfos to test shape analysis. By default, we just check that we don't give incorrect answers, and then if `assert_jit_shape_analysis` is true, tests that we correctly propagates the full shape. and it found a couple bugs 😃 [ghstack-poisoned]

ghstack-source-id: 1213d8a Pull Request resolved: #59814

eellison · 2021-08-09T16:42:47Z

@mruberry there may be changes on the other PR but I think this one is ready to go

Using opinfos to test shape analysis. By default, we just check that we don't give incorrect answers, and then if `assert_jit_shape_analysis` is true, tests that we correctly propagates the full shape. and it found a couple bugs 😃 [ghstack-poisoned]

eellison · 2021-08-09T19:57:24Z