profile ivalue for nvfuser by jjsjann123 · Pull Request #47668 · pytorch/pytorch

jjsjann123 · 2020-11-10T12:04:52Z

Stack from ghstack:

profile ivalue for nvfuser #47668 profile ivalue for nvfuser
refactor profiling optional #47667 refactor profiling optional

Differential Revision: D25255571

[ghstack-poisoned]

ghstack-source-id: 207164a Pull Request resolved: #47668

dr-ci · 2020-11-10T13:50:30Z

💊 CI failures summary and remediations

As of commit afbff83 (more details on the Dr. CI page):

9/9 failures possibly* introduced in this PR
- 2/9 non-CircleCI failure(s)

🕵️ 7 new failures recognized by patterns

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

pytorch_linux_bionic_py3_6_clang9_test (1/7)

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

Jan 13 08:29:40 AssertionError: False is not true :

Jan 13 08:29:40   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_utils.py", line 596, in wrapper
Jan 13 08:29:40     fn(*args, **kwargs)
Jan 13 08:29:40   File "test_jit.py", line 15749, in do_test
Jan 13 08:29:40     run_test()
Jan 13 08:29:40   File "test_jit.py", line 15743, in run_test
Jan 13 08:29:40     self.assertAutodiffNode(script_fn.last_graph, should_autodiff_node, autodiff_nodes, fusible_nodes)
Jan 13 08:29:40   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_jit.py", line 220, in assertAutodiffNode
Jan 13 08:29:40     found_all_nonfusible_nodes and found_all_fusible_nodes, err_msg)
Jan 13 08:29:40   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_utils.py", line 1216, in assertEqual
Jan 13 08:29:40     super().assertTrue(x == y, msg=msg)
Jan 13 08:29:40 AssertionError: False is not true : 
Jan 13 08:29:40 Failure in testing nodes' autodifferentiation, one or more nodes were expected to be autodiffed, but were not found in specified fusible/nonfusible DifferentiableGraph groups. 
Jan 13 08:29:40 Specifically:
Jan 13 08:29:40   aten::softmax was not in one of the DifferentiableGraphs when it was expected to be. Did you intend for this node to be autodiffed? If not, remove it from the list of nonfusible nodes.
Jan 13 08:29:40 
Jan 13 08:29:40 ----------------------------------------------------------------------
Jan 13 08:29:40 Ran 2887 tests in 166.777s
Jan 13 08:29:40 
Jan 13 08:29:40 FAILED (failures=84, errors=1, skipped=76, expected failures=1)
Jan 13 08:29:40 
Jan 13 08:29:40 Generating XML reports...

pytorch_linux_xenial_py3_6_gcc5_4_test (2/7)

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

Jan 13 08:31:49 AssertionError: False is not true :

Jan 13 08:31:49   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_utils.py", line 596, in wrapper
Jan 13 08:31:49     fn(*args, **kwargs)
Jan 13 08:31:49   File "test_jit.py", line 15749, in do_test
Jan 13 08:31:49     run_test()
Jan 13 08:31:49   File "test_jit.py", line 15743, in run_test
Jan 13 08:31:49     self.assertAutodiffNode(script_fn.last_graph, should_autodiff_node, autodiff_nodes, fusible_nodes)
Jan 13 08:31:49   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_jit.py", line 220, in assertAutodiffNode
Jan 13 08:31:49     found_all_nonfusible_nodes and found_all_fusible_nodes, err_msg)
Jan 13 08:31:49   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_utils.py", line 1216, in assertEqual
Jan 13 08:31:49     super().assertTrue(x == y, msg=msg)
Jan 13 08:31:49 AssertionError: False is not true : 
Jan 13 08:31:49 Failure in testing nodes' autodifferentiation, one or more nodes were expected to be autodiffed, but were not found in specified fusible/nonfusible DifferentiableGraph groups. 
Jan 13 08:31:49 Specifically:
Jan 13 08:31:49   aten::softmax was not in one of the DifferentiableGraphs when it was expected to be. Did you intend for this node to be autodiffed? If not, remove it from the list of nonfusible nodes.
Jan 13 08:31:49 
Jan 13 08:31:49 ----------------------------------------------------------------------
Jan 13 08:31:49 Ran 2887 tests in 151.740s
Jan 13 08:31:49 
Jan 13 08:31:49 FAILED (failures=84, errors=1, skipped=76, expected failures=1)
Jan 13 08:31:49 
Jan 13 08:31:49 Generating XML reports...

pytorch_linux_xenial_py3_clang5_asan_test2 (3/7)

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

Jan 13 08:43:32 AssertionError: False is not true :

Jan 13 08:43:32   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_utils.py", line 596, in wrapper
Jan 13 08:43:32     fn(*args, **kwargs)
Jan 13 08:43:32   File "test_jit.py", line 15749, in do_test
Jan 13 08:43:32     run_test()
Jan 13 08:43:32   File "test_jit.py", line 15743, in run_test
Jan 13 08:43:32     self.assertAutodiffNode(script_fn.last_graph, should_autodiff_node, autodiff_nodes, fusible_nodes)
Jan 13 08:43:32   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_jit.py", line 220, in assertAutodiffNode
Jan 13 08:43:32     found_all_nonfusible_nodes and found_all_fusible_nodes, err_msg)
Jan 13 08:43:32   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_utils.py", line 1216, in assertEqual
Jan 13 08:43:32     super().assertTrue(x == y, msg=msg)
Jan 13 08:43:32 AssertionError: False is not true : 
Jan 13 08:43:32 Failure in testing nodes' autodifferentiation, one or more nodes were expected to be autodiffed, but were not found in specified fusible/nonfusible DifferentiableGraph groups. 
Jan 13 08:43:32 Specifically:
Jan 13 08:43:32   aten::softmax was not in one of the DifferentiableGraphs when it was expected to be. Did you intend for this node to be autodiffed? If not, remove it from the list of nonfusible nodes.
Jan 13 08:43:32 
Jan 13 08:43:32 ----------------------------------------------------------------------
Jan 13 08:43:32 Ran 2856 tests in 739.654s
Jan 13 08:43:32 
Jan 13 08:43:32 FAILED (failures=84, errors=1, skipped=76, expected failures=1)
Jan 13 08:43:32 
Jan 13 08:43:32 Generating XML reports...

pytorch_macos_10_13_py3_test (4/7)

Step: "Test" (full log | diagnosis details | 🔁 rerun)

Jan 13 07:29:31 AssertionError: False is not true :

Jan 13 07:29:31   File "/Users/distiller/workspace/miniconda3/lib/python3.7/site-packages/torch/testing/_internal/common_utils.py", line 596, in wrapper
Jan 13 07:29:31     fn(*args, **kwargs)
Jan 13 07:29:31   File "test_jit.py", line 15749, in do_test
Jan 13 07:29:31     run_test()
Jan 13 07:29:31   File "test_jit.py", line 15743, in run_test
Jan 13 07:29:31     self.assertAutodiffNode(script_fn.last_graph, should_autodiff_node, autodiff_nodes, fusible_nodes)
Jan 13 07:29:31   File "/Users/distiller/workspace/miniconda3/lib/python3.7/site-packages/torch/testing/_internal/common_jit.py", line 220, in assertAutodiffNode
Jan 13 07:29:31     found_all_nonfusible_nodes and found_all_fusible_nodes, err_msg)
Jan 13 07:29:31   File "/Users/distiller/workspace/miniconda3/lib/python3.7/site-packages/torch/testing/_internal/common_utils.py", line 1216, in assertEqual
Jan 13 07:29:31     super().assertTrue(x == y, msg=msg)
Jan 13 07:29:31 AssertionError: False is not true : 
Jan 13 07:29:31 Failure in testing nodes' autodifferentiation, one or more nodes were expected to be autodiffed, but were not found in specified fusible/nonfusible DifferentiableGraph groups. 
Jan 13 07:29:31 Specifically:
Jan 13 07:29:31   aten::softmax was not in one of the DifferentiableGraphs when it was expected to be. Did you intend for this node to be autodiffed? If not, remove it from the list of nonfusible nodes.
Jan 13 07:29:31 
Jan 13 07:29:31 ----------------------------------------------------------------------
Jan 13 07:29:31 Ran 2887 tests in 199.141s
Jan 13 07:29:31 
Jan 13 07:29:31 FAILED (failures=84, errors=1, skipped=107, expected failures=1)
Jan 13 07:29:31 
Jan 13 07:29:31 Generating XML reports...

pytorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_test2 (5/7)

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

Jan 13 10:27:03 AssertionError: False is not true :

Jan 13 10:27:03   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_utils.py", line 596, in wrapper
Jan 13 10:27:03     fn(*args, **kwargs)
Jan 13 10:27:03   File "test_jit.py", line 15749, in do_test
Jan 13 10:27:03     run_test()
Jan 13 10:27:03   File "test_jit.py", line 15743, in run_test
Jan 13 10:27:03     self.assertAutodiffNode(script_fn.last_graph, should_autodiff_node, autodiff_nodes, fusible_nodes)
Jan 13 10:27:03   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_jit.py", line 220, in assertAutodiffNode
Jan 13 10:27:03     found_all_nonfusible_nodes and found_all_fusible_nodes, err_msg)
Jan 13 10:27:03   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_utils.py", line 1216, in assertEqual
Jan 13 10:27:03     super().assertTrue(x == y, msg=msg)
Jan 13 10:27:03 AssertionError: False is not true : 
Jan 13 10:27:03 Failure in testing nodes' autodifferentiation, one or more nodes were expected to be autodiffed, but were not found in specified fusible/nonfusible DifferentiableGraph groups. 
Jan 13 10:27:03 Specifically:
Jan 13 10:27:03   aten::softmax was not in one of the DifferentiableGraphs when it was expected to be. Did you intend for this node to be autodiffed? If not, remove it from the list of nonfusible nodes.
Jan 13 10:27:03 
Jan 13 10:27:03 ----------------------------------------------------------------------
Jan 13 10:27:03 Ran 2889 tests in 209.434s
Jan 13 10:27:03 
Jan 13 10:27:03 FAILED (failures=73, errors=1, skipped=43, expected failures=1)
Jan 13 10:27:03 
Jan 13 10:27:03 Generating XML reports...

pytorch_linux_bionic_py3_8_gcc9_coverage_test2 (6/7)

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

Jan 13 08:13:20 AssertionError: False is not true :

Jan 13 08:13:20   File "/opt/conda/lib/python3.8/site-packages/torch/testing/_internal/common_utils.py", line 596, in wrapper
Jan 13 08:13:20     fn(*args, **kwargs)
Jan 13 08:13:20   File "test_jit.py", line 15749, in do_test
Jan 13 08:13:20     run_test()
Jan 13 08:13:20   File "test_jit.py", line 15743, in run_test
Jan 13 08:13:20     self.assertAutodiffNode(script_fn.last_graph, should_autodiff_node, autodiff_nodes, fusible_nodes)
Jan 13 08:13:20   File "/opt/conda/lib/python3.8/site-packages/torch/testing/_internal/common_jit.py", line 219, in assertAutodiffNode
Jan 13 08:13:20     self.assertEqual(should_autodiff_node,
Jan 13 08:13:20   File "/opt/conda/lib/python3.8/site-packages/torch/testing/_internal/common_utils.py", line 1216, in assertEqual
Jan 13 08:13:20     super().assertTrue(x == y, msg=msg)
Jan 13 08:13:20 AssertionError: False is not true : 
Jan 13 08:13:20 Failure in testing nodes' autodifferentiation, one or more nodes were expected to be autodiffed, but were not found in specified fusible/nonfusible DifferentiableGraph groups. 
Jan 13 08:13:20 Specifically:
Jan 13 08:13:20   aten::softmax was not in one of the DifferentiableGraphs when it was expected to be. Did you intend for this node to be autodiffed? If not, remove it from the list of nonfusible nodes.
Jan 13 08:13:20 
Jan 13 08:13:21 ----------------------------------------------------------------------
Jan 13 08:13:21 Ran 2887 tests in 361.683s
Jan 13 08:13:21 
Jan 13 08:13:21 FAILED (failures=84, errors=1, skipped=76, expected failures=1)
Jan 13 08:13:21 
Jan 13 08:13:21 Generating XML reports...

pytorch_windows_vs2019_py36_cuda10.1_test2 (7/7)

Step: "Test" (full log | diagnosis details | 🔁 rerun)

AssertionError: False is not true :

  File "C:\Users\circleci\project\build\win_tmp\build\torch\testing\_internal\common_utils.py", line 596, in wrapper
    fn(*args, **kwargs)
  File "test_jit.py", line 15749, in do_test
    run_test()
  File "test_jit.py", line 15743, in run_test
    self.assertAutodiffNode(script_fn.last_graph, should_autodiff_node, autodiff_nodes, fusible_nodes)
  File "C:\Users\circleci\project\build\win_tmp\build\torch\testing\_internal\common_jit.py", line 220, in assertAutodiffNode
    found_all_nonfusible_nodes and found_all_fusible_nodes, err_msg)
  File "C:\Users\circleci\project\build\win_tmp\build\torch\testing\_internal\common_utils.py", line 1216, in assertEqual
    super().assertTrue(x == y, msg=msg)
AssertionError: False is not true : 
Failure in testing nodes' autodifferentiation, one or more nodes were expected to be autodiffed, but were not found in specified fusible/nonfusible DifferentiableGraph groups. 
Specifically:
  aten::softmax was not in one of the DifferentiableGraphs when it was expected to be. Did you intend for this node to be autodiffed? If not, remove it from the list of nonfusible nodes.

----------------------------------------------------------------------
Ran 2889 tests in 177.091s

FAILED (failures=73, errors=1, skipped=82, expected failures=1)

Generating XML reports...

Extra GitHub checks: 1 failed

Failed: GitHub Actions - clang-format

ci.pytorch.org: 1 failed

Failed: pr/pytorch-linux-bionic-rocm3.10-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.

[ghstack-poisoned]

ghstack-source-id: 9d2d2c2 Pull Request resolved: #47668

eellison · 2020-12-02T23:41:50Z

@jjsjann123 can you get someone from nvidia to review and accept before we land this ? @Krovatkin

tlemo · 2020-12-03T17:54:36Z

+void profileIntList(ProfilingRecord* pr, Node* node, size_t offset) {
+  auto pn = insertProfileIValueOp(node, offset, pr);
+
+  std::function<void(Stack&)> ivalue_profiler = [pr, pn](Stack& stack) {


why assign to a std::function here? (rather than say ... const auto)

tlemo · 2020-12-03T17:54:56Z

+void profileBool(ProfilingRecord* pr, Node* node, size_t offset) {
+  auto pn = insertProfileIValueOp(node, offset, pr);
+
+  std::function<void(Stack&)> ivalue_profiler = [pr, pn](Stack& stack) {


tlemo · 2020-12-03T17:56:12Z

  return IrParser::canParseNode(node);
 }

+// TODO: we should incorporate this to our parser as well;


It's not clear what our parser refers to in this context: replace our parser to a more explicit pointer ?

tlemo · 2020-12-03T17:57:06Z

+          ->schema();
+  if (node->matches(reduction_operator_schema)) {
+    switch (offset) {
+      case 1:


what do 1 and 2 mean here?

tlemo · 2020-12-03T17:57:56Z

+}
+
+void insertProfileNodesForCUDAFuser_(Block* block, ProfilingRecord* pr) {
+  for (auto it = block->nodes().begin(); it != block->nodes().end(); ++it) {


for (const auto& n : block->nodes()) ?

tlemo · 2020-12-03T17:59:46Z


    // TODO: failure in buildShapeExpressions should not break fusion execution,
    // we can add a try/catch here to bailout from removeOutputsUsedOnlyInSize.
+    GRAPH_DUMP("before build shape expression: ", graph_);


does this unconditionally dump the graph?

Nope, this is only enabled via environment variable:

pytorch/torch/csrc/jit/jit_log.h

Lines 6 to 37 in cbc34db

// `TorchScript` offers a simple logging facility that can enabled by setting an

// environment variable `PYTORCH_JIT_LOG_LEVEL`.

// Logging is enabled on a per file basis. To enable logging in

// `dead_code_elimination.cpp`, `PYTORCH_JIT_LOG_LEVEL` should be

// set to `dead_code_elimination.cpp` or, simply, to `dead_code_elimination`

// (i.e. `PYTORCH_JIT_LOG_LEVEL=dead_code_elimination`).

// Multiple files can be logged by separating each file name with a colon `:` as

// in the following example,

// `PYTORCH_JIT_LOG_LEVEL=dead_code_elimination:guard_elimination`

// There are 3 logging levels available for your use ordered by the detail level

// from lowest to highest.

// * `GRAPH_DUMP` should be used for printing entire graphs after optimization

// passes

// * `GRAPH_UPDATE` should be used for reporting graph transformations (i.e.

// node deletion, constant folding, etc)

// * `GRAPH_DEBUG` should be used for providing information useful for debugging

// the internals of a particular optimization pass or analysis

// The default logging level is `GRAPH_DUMP` meaning that only `GRAPH_DUMP`

// statements will be enabled when one specifies a file(s) in

// `PYTORCH_JIT_LOG_LEVEL`.

// `GRAPH_UPDATE` can be enabled by prefixing a file name with an `>` as in

// `>alias_analysis`.

// `GRAPH_DEBUG` can be enabled by prefixing a file name with an `>>` as in

// `>>alias_analysis`.

// `>>>` is also valid and **currently** is equivalent to `GRAPH_DEBUG` as there

// is no logging level that is higher than `GRAPH_DEBUG`.

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

ghstack-source-id: 4b784cb Pull Request resolved: #47668

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

ghstack-source-id: c9106c9 Pull Request resolved: #47668

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

ghstack-source-id: 129a345 Pull Request resolved: #47668

…or nvfuser. We tried to go around PR pytorch#47667 refactor profiling optional, since upstream is still working on it at this time.

* This is a cherry-pick from upstream PR pytorch#47668 profile ivalue for nvfuser. We tried to go around PR pytorch#47667 refactor profiling optional, since upstream is still working on it at this time. createConditionalConstant supports profile ivalue including bool, int_list and size New guard to check conditional constant at runtime size_eq_guard op to facilitate comparison of dynamic sizes sum_to_size & _grad_sum_to_size added in integration

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

ghstack-source-id: 29c6d7a Pull Request resolved: #47668

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

ghstack-source-id: 5a3aa53 Pull Request resolved: #47668

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

ghstack-source-id: b0654c7 Pull Request resolved: #47668

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

ghstack-source-id: 095feb5 Pull Request resolved: #47668

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

ghstack-source-id: f6203c7 Pull Request resolved: #47668

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

ghstack-source-id: 343bbfc Pull Request resolved: #47668

profile ivalue for nvfuser

0dc58a1

[ghstack-poisoned]

jjsjann123 requested a review from apaszke as a code owner November 10, 2020 12:04

This was referenced Nov 10, 2020

Remove output used only by sizes (#448) #47665

Closed

adding profile_ivalue #47666

Closed

refactor profiling optional #47667

Closed

jjsjann123 added a commit that referenced this pull request Nov 10, 2020

profile ivalue for nvfuser

4f5aeb4

ghstack-source-id: 207164a Pull Request resolved: #47668

facebook-github-bot added cla signed oncall: jit Add this issue/PR to JIT oncall triage queue labels Nov 10, 2020

pytorchbot added the open source label Nov 10, 2020

kevinstephano mentioned this pull request Nov 16, 2020

Tracking of Fuser Issues for HuggingFace Bert Model csarofeen/pytorch#521

Closed

Update on "profile ivalue for nvfuser"

14d781c

[ghstack-poisoned]

Update on "profile ivalue for nvfuser"

cbc34db

[ghstack-poisoned]

jjsjann123 added a commit that referenced this pull request Nov 20, 2020

profile ivalue for nvfuser

294607a

ghstack-source-id: 9d2d2c2 Pull Request resolved: #47668

jjsjann123 requested a review from csarofeen December 3, 2020 14:10

tlemo reviewed Dec 3, 2020

View reviewed changes

Update on "profile ivalue for nvfuser"

01fbae1

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

Update on "profile ivalue for nvfuser"

29e201e

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

jjsjann123 added a commit that referenced this pull request Dec 3, 2020

profile ivalue for nvfuser

b719dea

ghstack-source-id: 4b784cb Pull Request resolved: #47668

Update on "profile ivalue for nvfuser"

eb66368

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

jjsjann123 added a commit that referenced this pull request Dec 8, 2020

profile ivalue for nvfuser

5f32d72

ghstack-source-id: c9106c9 Pull Request resolved: #47668

Update on "profile ivalue for nvfuser"

2f46f5d

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

jjsjann123 added a commit that referenced this pull request Dec 8, 2020

profile ivalue for nvfuser

f8bad73

ghstack-source-id: 129a345 Pull Request resolved: #47668

Update on "profile ivalue for nvfuser"

6979391

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

jjsjann123 added a commit that referenced this pull request Jan 12, 2021

profile ivalue for nvfuser

2183ffd

ghstack-source-id: 29c6d7a Pull Request resolved: #47668

Update on "profile ivalue for nvfuser"

7497ed8

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

jjsjann123 added a commit that referenced this pull request Jan 12, 2021

profile ivalue for nvfuser

f49bd41

ghstack-source-id: 5a3aa53 Pull Request resolved: #47668

Update on "profile ivalue for nvfuser"

b39b7e8

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

Update on "profile ivalue for nvfuser"

afbff83

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

jjsjann123 added a commit that referenced this pull request Jan 13, 2021

profile ivalue for nvfuser

57c198e

ghstack-source-id: b0654c7 Pull Request resolved: #47668

Update on "profile ivalue for nvfuser"

1196cc4

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

Update on "profile ivalue for nvfuser"

8e3342d

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

jjsjann123 added a commit that referenced this pull request Jan 16, 2021

profile ivalue for nvfuser

2e2a3e1

ghstack-source-id: 095feb5 Pull Request resolved: #47668

Update on "profile ivalue for nvfuser"

c30e03f

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

jjsjann123 added a commit that referenced this pull request Jan 19, 2021

profile ivalue for nvfuser

18d852e

ghstack-source-id: f6203c7 Pull Request resolved: #47668

Update on "profile ivalue for nvfuser"

e6f11cf

Differential Revision: [D25255571](https://our.internmc.facebook.com/intern/diff/D25255571) [ghstack-poisoned]

jjsjann123 added a commit that referenced this pull request Jan 20, 2021

profile ivalue for nvfuser

dac168d

ghstack-source-id: 343bbfc Pull Request resolved: #47668

jjsjann123 closed this Jan 29, 2021

facebook-github-bot deleted the gh/jjsjann123/4/head branch March 1, 2021 15:16

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

profile ivalue for nvfuser#47668

profile ivalue for nvfuser#47668
jjsjann123 wants to merge 15 commits intogh/jjsjann123/4/basefrom
gh/jjsjann123/4/head

jjsjann123 commented Nov 10, 2020 •

edited

Loading

Uh oh!

dr-ci Bot commented Nov 10, 2020 •

edited by facebook-github-bot

Loading

Uh oh!

eellison commented Dec 2, 2020

Uh oh!

tlemo Dec 3, 2020

Uh oh!

tlemo Dec 3, 2020

Uh oh!

tlemo Dec 3, 2020

Uh oh!

tlemo Dec 3, 2020

Uh oh!

tlemo Dec 3, 2020

Uh oh!

tlemo Dec 3, 2020

Uh oh!

jjsjann123 Dec 28, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

	// `TorchScript` offers a simple logging facility that can enabled by setting an
	// environment variable `PYTORCH_JIT_LOG_LEVEL`.

	// Logging is enabled on a per file basis. To enable logging in
	// `dead_code_elimination.cpp`, `PYTORCH_JIT_LOG_LEVEL` should be
	// set to `dead_code_elimination.cpp` or, simply, to `dead_code_elimination`
	// (i.e. `PYTORCH_JIT_LOG_LEVEL=dead_code_elimination`).

	// Multiple files can be logged by separating each file name with a colon `:` as
	// in the following example,
	// `PYTORCH_JIT_LOG_LEVEL=dead_code_elimination:guard_elimination`

	// There are 3 logging levels available for your use ordered by the detail level
	// from lowest to highest.

	// * `GRAPH_DUMP` should be used for printing entire graphs after optimization
	// passes
	// * `GRAPH_UPDATE` should be used for reporting graph transformations (i.e.
	// node deletion, constant folding, etc)
	// * `GRAPH_DEBUG` should be used for providing information useful for debugging
	// the internals of a particular optimization pass or analysis

	// The default logging level is `GRAPH_DUMP` meaning that only `GRAPH_DUMP`
	// statements will be enabled when one specifies a file(s) in
	// `PYTORCH_JIT_LOG_LEVEL`.

	// `GRAPH_UPDATE` can be enabled by prefixing a file name with an `>` as in
	// `>alias_analysis`.
	// `GRAPH_DEBUG` can be enabled by prefixing a file name with an `>>` as in
	// `>>alias_analysis`.
	// `>>>` is also valid and currently is equivalent to `GRAPH_DEBUG` as there
	// is no logging level that is higher than `GRAPH_DEBUG`.

Conversation

jjsjann123 commented Nov 10, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dr-ci Bot commented Nov 10, 2020 • edited by facebook-github-bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

💊 CI failures summary and remediations

🕵️ 7 new failures recognized by patterns

pytorch_linux_bionic_py3_6_clang9_test (1/7)

pytorch_linux_xenial_py3_6_gcc5_4_test (2/7)

pytorch_linux_xenial_py3_clang5_asan_test2 (3/7)

pytorch_macos_10_13_py3_test (4/7)

pytorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_test2 (5/7)

pytorch_linux_bionic_py3_8_gcc9_coverage_test2 (6/7)

pytorch_windows_vs2019_py36_cuda10.1_test2 (7/7)

Extra GitHub checks: 1 failed

ci.pytorch.org: 1 failed

Uh oh!

eellison commented Dec 2, 2020

Uh oh!

tlemo Dec 3, 2020

Choose a reason for hiding this comment

Uh oh!

tlemo Dec 3, 2020

Choose a reason for hiding this comment

Uh oh!

tlemo Dec 3, 2020

Choose a reason for hiding this comment

Uh oh!

tlemo Dec 3, 2020

Choose a reason for hiding this comment

Uh oh!

tlemo Dec 3, 2020

Choose a reason for hiding this comment

Uh oh!

tlemo Dec 3, 2020

Choose a reason for hiding this comment

Uh oh!

jjsjann123 Dec 28, 2020

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

jjsjann123 commented Nov 10, 2020 •

edited

Loading

dr-ci Bot commented Nov 10, 2020 •

edited by facebook-github-bot

Loading