Rocm ep acc test support by groenenboomj · Pull Request #13306 · microsoft/onnxruntime

groenenboomj · 2022-10-13T06:14:55Z

Description

Add support for ROCm EP in accuracy tests

Motivation and Context

This allows ONNXRT accuracy tests to be ran on the ROCm EP.

cloudhan · 2022-10-14T04:54:11Z

    import onnxruntime

-    if use_gpu and "CUDAExecutionProvider" not in onnxruntime.get_available_providers():
+    if use_gpu and "CUDAExecutionProvider" and "ROCMExecutionProvider" not in onnxruntime.get_available_providers():


address py lint "CUDAExecutionProvider" and "ROCMExecutionProvider"

cloudhan · 2022-10-14T04:57:48Z

+        assert "CUDAExecutionProvider" or "ROCMExecutionProvider" in session.get_providers()  # Make sure there is GPU

    assert os.path.exists(optimized_model_path) and os.path.isfile(optimized_model_path)
    logger.debug("Save optimized model by onnxruntime to {}".format(optimized_model_path))


py lint. This can be changed to logger.debug("Save optimized model by onnxruntime to %s", optimized_model_path) to delay the string formation.

lgtm-com · 2022-10-20T23:29:24Z

This pull request fixes 1 alert when merging 2b6e010 into 7a3486c - view on LGTM.com

fixed alerts:

1 for Unused local variable

lgtm-com · 2022-10-21T22:52:39Z

This pull request fixes 1 alert when merging 8561cd8 into 928c988 - view on LGTM.com

fixed alerts:

1 for Unused local variable

tianleiwu · 2022-10-24T20:41:33Z

+            onnx_model_path, sess_options, providers=gpu_ep, **kwargs
        )
-        assert "CUDAExecutionProvider" in session.get_providers()  # Make sure there is GPU
+        assert any(gpu_ep in session.get_providers() for gpu_ep in ["CUDAExecutionProvider","ROCMExecutionProvider"]) # Make sure there is GPU


Remove this line since I think session creation will fail if no CUDA or Rocm EP, and line 76 has checked.

Tested and removed. I agree.

Update benchmark_helper.py to support ROCm back end.

lgtm-com · 2022-11-02T22:02:58Z

This pull request introduces 1 alert when merging 5c4e255 into b1e1b25 - view on LGTM.com

new alerts:

1 for Unused local variable

tianleiwu · 2022-11-18T20:11:22Z

/azp run Windows CPU CI Pipeline,Windows GPU CI Pipeline,Windows GPU TensorRT CI Pipeline,ONNX Runtime Web CI Pipeline,Linux CPU CI Pipeline,Linux CPU Minimal Build E2E CI Pipeline,Linux GPU CI Pipeline,Linux GPU TensorRT CI Pipeline,Linux OpenVINO CI Pipeline,MacOS CI Pipeline

tianleiwu · 2022-11-18T20:11:23Z

/azp run orttraining-amd-gpu-ci-pipeline,orttraining-linux-ci-pipeline,orttraining-linux-gpu-ci-pipeline,orttraining-ortmodule-distributed,onnxruntime-python-checks-ci-pipeline,onnxruntime-binary-size-checks-ci-pipeline

azure-pipelines · 2022-11-18T20:11:49Z

Azure Pipelines successfully started running 6 pipeline(s).

azure-pipelines · 2022-11-18T20:11:58Z

Azure Pipelines successfully started running 10 pipeline(s).

Add GPU environment check.

lgtm-com · 2022-11-18T21:39:57Z

This pull request fixes 1 alert when merging 75ac09b into 3e9e5e9 - view on LGTM.com

fixed alerts:

1 for Unused local variable

Heads-up: LGTM.com's PR analysis will be disabled on the 5th of December, and LGTM.com will be shut down ⏻ completely on the 16th of December 2022. It looks like GitHub code scanning with CodeQL is already set up for this repo, so no further action is needed 🚀. For more information, please check out our post on the GitHub blog.

lgtm-com · 2022-11-18T22:28:19Z

This pull request fixes 1 alert when merging 2461ad6 into 3e9e5e9 - view on LGTM.com

fixed alerts:

1 for Unused local variable

Heads-up: LGTM.com's PR analysis will be disabled on the 5th of December, and LGTM.com will be shut down ⏻ completely on the 16th of December 2022. It looks like GitHub code scanning with CodeQL is already set up for this repo, so no further action is needed 🚀. For more information, please check out our post on the GitHub blog.

cloudhan · 2022-11-21T12:30:43Z

-            assert (
-                "CUDAExecutionProvider" in onnxruntime.get_available_providers()
-            ), "Please install onnxruntime-gpu package to test GPU inference."
+            assert any(


Please use set disjoint test for readibility:

set(onnxruntime.get_available_providers()).isdisjoint(["CUDAExecutionProvider", "ROCMExecutionProvider"])

cloudhan · 2022-11-21T12:30:58Z

    import onnxruntime

-    if use_gpu and "CUDAExecutionProvider" not in onnxruntime.get_available_providers():
+    if use_gpu and not any(


cloudhan · 2022-11-21T12:31:07Z

+            gpu_ep.append("ROCMExecutionProvider")
+
+        session = onnxruntime.InferenceSession(onnx_model_path, sess_options, providers=gpu_ep, **kwargs)
+        assert any(


Allows MIGraphX EP to run the following additional tests. Also adds support to get MIGraphX to run eval_squad.py Reference to the Rocm EP changes: #13306 Co-authored-by: Joseph Groenenboom <joseph.groenenboom@amd.com> Co-authored-by: Ted Themistokleous <tthemist@amd.com>

groenenboomj · 2022-12-02T20:33:55Z

The changes in this PR were rolled into and merged with #13455. Closing.

Allows MIGraphX EP to run the following additional tests. Also adds support to get MIGraphX to run eval_squad.py Reference to the Rocm EP changes: microsoft#13306 Co-authored-by: Joseph Groenenboom <joseph.groenenboom@amd.com> Co-authored-by: Ted Themistokleous <tthemist@amd.com>

cloudhan reviewed Oct 14, 2022

View reviewed changes

Comment thread onnxruntime/python/tools/transformers/optimizer.py Outdated

cloudhan reviewed Oct 14, 2022

View reviewed changes

tianleiwu reviewed Oct 24, 2022

View reviewed changes

TedThemistokleous mentioned this pull request Oct 26, 2022

Modify MIGraphX EP for Accuracy tests #13455

Merged

groenenboomj added 7 commits November 2, 2022 11:22

Add support for ROCm EP in benchmark tests.

bdeabd5

Update benchmark_helper.py to support ROCm back end.

Add more ROCm support for test cases

de6e4f7

Clean up GPU Ep assignment and checks

608cfbb

Address pylint and update GPU check in session

7a03ee2

Add check for GPU backend in optimizer.py

7ca4805

Remove redundant check for GPU backend

755a206

Add ROCm EP support to eval_squad.py

5c4e255

groenenboomj force-pushed the rocm_ep_acc_test_support branch from 8561cd8 to 5c4e255 Compare November 2, 2022 21:46

tianleiwu previously approved these changes Nov 18, 2022

View reviewed changes

groenenboomj marked this pull request as ready for review November 18, 2022 16:59

Python linting and formatting

75ac09b

Add GPU environment check.

groenenboomj dismissed tianleiwu’s stale review via 75ac09b November 18, 2022 21:19

This comment was marked as resolved.

Sign in to view

Update import ordering

2461ad6

cloudhan approved these changes Nov 21, 2022

View reviewed changes

This comment was marked as resolved.

Sign in to view

cloudhan reviewed Nov 21, 2022

View reviewed changes

tianleiwu approved these changes Nov 21, 2022

View reviewed changes

groenenboomj closed this Dec 2, 2022

Conversation

groenenboomj commented Oct 13, 2022

Description

Motivation and Context

Uh oh!

cloudhan Oct 14, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

cloudhan Oct 14, 2022

Choose a reason for hiding this comment

Uh oh!

lgtm-com Bot commented Oct 20, 2022

Uh oh!

lgtm-com Bot commented Oct 21, 2022

Uh oh!

tianleiwu Oct 24, 2022

Choose a reason for hiding this comment

Uh oh!

groenenboomj Nov 2, 2022

Choose a reason for hiding this comment

Uh oh!

lgtm-com Bot commented Nov 2, 2022

Uh oh!

tianleiwu commented Nov 18, 2022

Uh oh!

tianleiwu commented Nov 18, 2022

Uh oh!

azure-pipelines Bot commented Nov 18, 2022

Uh oh!

azure-pipelines Bot commented Nov 18, 2022

Uh oh!

This comment was marked as resolved.

This comment was marked as resolved.

This comment was marked as resolved.

This comment was marked as resolved.

lgtm-com Bot commented Nov 18, 2022

Uh oh!

lgtm-com Bot commented Nov 18, 2022

Uh oh!

This comment was marked as resolved.

This comment was marked as resolved.

This comment was marked as resolved.

This comment was marked as resolved.

This comment was marked as resolved.

This comment was marked as resolved.

This comment was marked as resolved.

cloudhan Nov 21, 2022

Choose a reason for hiding this comment

Uh oh!

cloudhan Nov 21, 2022

Choose a reason for hiding this comment

Uh oh!

cloudhan Nov 21, 2022

Choose a reason for hiding this comment

Uh oh!

groenenboomj commented Dec 2, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

cloudhan Oct 14, 2022 •

edited

Loading