ns for fx: support binary ops when adding unshadowed loggers for inputs

[vkuzo]

Stack from ghstack:

• ns for fx: additional bugfix for user defined functions #57028 ns for fx: additional bugfix for user defined functions
• ns for fx: allow comparing int8 to int8 for functionals #57027 ns for fx: allow comparing int8 to int8 for functionals
• ns for fx: add option to skip matching classes and functions #57026 ns for fx: add option to skip matching classes and functions
• ns for fx: support binary ops when adding unshadowed loggers for inputs #57025 ns for fx: support binary ops when adding unshadowed loggers for inputs
• ns for fx: bug fix for shadowing fp16 emulation patterns #57024 ns for fx: bug fix for shadowing fp16 emulation patterns
• ns for fx: add fp16 function shadowing #57023 ns for fx: add fp16 function shadowing
• ns for fx: allow user functions in shadowing #57022 ns for fx: allow user functions in shadowing
• ns for fx: move node I/O dtype mapping to be local instead of global #57021 ns for fx: move node I/O dtype mapping to be local instead of global

Summary:

Adds the ability to log unshadowed inputs of binary ops such as add
and mul, when indices 0, 1, or 0 and 1 are tensors.

Note: making shadowing support this is saved for a future PR.

Test Plan:

python test/test_quantization.py TestFXNumericSuiteCoreAPIs.test_add_mul_inputs_activations

Differential Revision: D28030098

[facebook-github-bot]

💊 CI failures summary and remediations

As of commit 131c00c (more details on the Dr. CI page):

---

✅ None of the CI failures appear to be your fault 💚

---

• 1/1 tentatively recognized as flaky ❄️
  • Click here to rerun these jobs

---

❄️ 1 failure tentatively classified as flaky

but reruns have not yet been triggered to confirm:

pytorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_test2 (1/1)

Step: "Run tests" (full log | diagnosis details | 🔁 rerun) ❄️

Apr 27 16:57:12 RuntimeError: Process 0 terminated or timed out after 110.04590153694153 seconds

Apr 27 16:57:12 ======================================================================
Apr 27 16:57:12 ERROR [110.069s]: test_nccl_high_priority_stream (__main__.TestDistBackendWithSpawn)
Apr 27 16:57:12 ----------------------------------------------------------------------
Apr 27 16:57:12 Traceback (most recent call last):
Apr 27 16:57:12   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_distributed.py", line 374, in wrapper
Apr 27 16:57:12     self._join_processes(fn)
Apr 27 16:57:12   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_distributed.py", line 566, in _join_processes
Apr 27 16:57:12     self._check_return_codes(elapsed_time)
Apr 27 16:57:12   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_distributed.py", line 614, in _check_return_codes
Apr 27 16:57:12     raise RuntimeError('Process {} terminated or timed out after {} seconds'.format(i, elapsed_time))
Apr 27 16:57:12 RuntimeError: Process 0 terminated or timed out after 110.04590153694153 seconds
Apr 27 16:57:12 
Apr 27 16:57:12 ----------------------------------------------------------------------
Apr 27 16:57:12 Ran 196 tests in 658.033s
Apr 27 16:57:12 
Apr 27 16:57:12 FAILED (errors=4, skipped=117)
Apr 27 16:57:12 
Apr 27 16:57:12 Generating XML reports...
Apr 27 16:57:12 Generated XML report: test-reports/dist-nccl/distributed.test_distributed_spawn/TEST-TestDistBackendWithSpawn-20210427164614.xml
Apr 27 16:57:12 Traceback (most recent call last):
Apr 27 16:57:12   File "test/run_test.py", line 1156, in <module>

---

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.

[facebook-github-bot]

@vkuzo merged this pull request in e8a5490.