Disable MatMulIntegerToFloat transformation for FP16 on CPU EP #18239
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
yufenglee
approved these changes
Nov 2, 2023
Contributor
Author
sumitsays
reviewed
Nov 2, 2023
Contributor
|
We should also add a test case in |
Contributor
It looks like we can manually assign each node to DML ep to test the fusion happens for DML EP in the case FP16 before calling the ApplyTransformers. for (auto& node : graph.Nodes()) {
node.SetExecutionProviderType("kDmlExecutionProvider");
}
And then we would be able to assert the count of MatMulIntegerToFloat node. |
sumitsays
reviewed
Nov 3, 2023
sumitsays
reviewed
Nov 3, 2023
sumitsays
reviewed
Nov 3, 2023
sumitsays
reviewed
Nov 3, 2023
AnaghaRaoAMD
force-pushed
the
user/anagrao/MatMulupdate
branch
from
94de302 to
cdb94fa
Compare
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| if __name__ == "__main__": | ||
| GenerateModel("matmul_integer_to_float.onnx") | ||
| GenerateModel("matmul_integer_to_float16.onnx", output_type_fp16=True) | ||
| GenerateModel("matmul_integer_to_float.onnx") |
Check warning
Code scanning / lintrunner
RUFF/W292
AnaghaRaoAMD
added a commit
that referenced
this pull request
Nov 3, 2023
### Description MatMulIntegerToFloat is updated to support FP16. The nodes for FP16 Transformation use "Mul" FP16, which is not directly supported by the CPU. For now FP16 transformation is only supported for DML EP. Disabled all FP16 tests on CPU. Tests result without `-use_dml` build flag ``` onnxruntime_test_all.exe --gtest_filter="*MatMulIntegerToFloat*" Note: Google Test filter = *MatMulIntegerToFloat* [==========] Running 8 tests from 4 test suites. [----------] Global test environment set-up. [----------] 1 test from CPU_U8S8_Precision_Tests [ RUN ] CPU_U8S8_Precision_Tests.MatMulIntegerToFloat [ OK ] CPU_U8S8_Precision_Tests.MatMulIntegerToFloat (181 ms) [----------] 1 test from CPU_U8S8_Precision_Tests (181 ms total) [----------] 1 test from GraphTransformationTests [ RUN ] GraphTransformationTests.MatMulIntegerToFloatTest [ OK ] GraphTransformationTests.MatMulIntegerToFloatTest (17 ms) [----------] 1 test from GraphTransformationTests (17 ms total) [----------] 1 test from QDQTransformerTests [ RUN ] QDQTransformerTests.MatMulIntegerToFloat [ OK ] QDQTransformerTests.MatMulIntegerToFloat (656 ms) [----------] 1 test from QDQTransformerTests (656 ms total) [----------] 5 tests from MatMulIntegerToFloat [ RUN ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_U8X8 [ OK ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_U8X8 (195 ms) [ RUN ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_U8X8 [ OK ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_U8X8 (206 ms) [ RUN ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_S8S8 [ OK ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_S8S8 (107 ms) [ RUN ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_S8S8 [ OK ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_S8S8 (114 ms) [ RUN ] MatMulIntegerToFloat.MatMulInteger_With_ZeroPoint [ OK ] MatMulIntegerToFloat.MatMulInteger_With_ZeroPoint (227 ms) [----------] 5 tests from MatMulIntegerToFloat (854 ms total) [----------] Global test environment tear-down [==========] 8 tests from 4 test suites ran. (1713 ms total) [ PASSED ] 8 tests. memleakdbg: ----- No memory leaks detected ----- ``` ``` onnxruntime_test_all.exe --gtest_filter="GraphTransformationTests.MatMulIntegerToFloat*" Note: Google Test filter = GraphTransformationTests.MatMulIntegerToFloat* [==========] Running 2 tests from 1 test suite. [----------] Global test environment set-up. [----------] 2 tests from GraphTransformationTests [ RUN ] GraphTransformationTests.MatMulIntegerToFloatTest [ OK ] GraphTransformationTests.MatMulIntegerToFloatTest (13 ms) [ RUN ] GraphTransformationTests.MatMulIntegerToFloat16Test [ OK ] GraphTransformationTests.MatMulIntegerToFloat16Test (4 ms) [----------] 2 tests from GraphTransformationTests (20 ms total) [----------] Global test environment tear-down [==========] 2 tests from 1 test suite ran. (22 ms total) [ PASSED ] 2 tests. memleakdbg: ----- No memory leaks detected ----- ``` ### Motivation and Context <!-- - Why is this change required? What problem does it solve? - If it fixes an open issue, please link to the issue here. -->
AnaghaRaoAMD
added a commit
that referenced
this pull request
Nov 7, 2023
### Description MatMulIntegerToFloat is updated to support FP16. The nodes for FP16 Transformation use "Mul" FP16, which is not directly supported by the CPU. For now FP16 transformation is only supported for DML EP. Disabled all FP16 tests on CPU. Tests result without `-use_dml` build flag ``` onnxruntime_test_all.exe --gtest_filter="*MatMulIntegerToFloat*" Note: Google Test filter = *MatMulIntegerToFloat* [==========] Running 8 tests from 4 test suites. [----------] Global test environment set-up. [----------] 1 test from CPU_U8S8_Precision_Tests [ RUN ] CPU_U8S8_Precision_Tests.MatMulIntegerToFloat [ OK ] CPU_U8S8_Precision_Tests.MatMulIntegerToFloat (181 ms) [----------] 1 test from CPU_U8S8_Precision_Tests (181 ms total) [----------] 1 test from GraphTransformationTests [ RUN ] GraphTransformationTests.MatMulIntegerToFloatTest [ OK ] GraphTransformationTests.MatMulIntegerToFloatTest (17 ms) [----------] 1 test from GraphTransformationTests (17 ms total) [----------] 1 test from QDQTransformerTests [ RUN ] QDQTransformerTests.MatMulIntegerToFloat [ OK ] QDQTransformerTests.MatMulIntegerToFloat (656 ms) [----------] 1 test from QDQTransformerTests (656 ms total) [----------] 5 tests from MatMulIntegerToFloat [ RUN ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_U8X8 [ OK ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_U8X8 (195 ms) [ RUN ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_U8X8 [ OK ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_U8X8 (206 ms) [ RUN ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_S8S8 [ OK ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_S8S8 (107 ms) [ RUN ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_S8S8 [ OK ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_S8S8 (114 ms) [ RUN ] MatMulIntegerToFloat.MatMulInteger_With_ZeroPoint [ OK ] MatMulIntegerToFloat.MatMulInteger_With_ZeroPoint (227 ms) [----------] 5 tests from MatMulIntegerToFloat (854 ms total) [----------] Global test environment tear-down [==========] 8 tests from 4 test suites ran. (1713 ms total) [ PASSED ] 8 tests. memleakdbg: ----- No memory leaks detected ----- ``` ``` onnxruntime_test_all.exe --gtest_filter="GraphTransformationTests.MatMulIntegerToFloat*" Note: Google Test filter = GraphTransformationTests.MatMulIntegerToFloat* [==========] Running 2 tests from 1 test suite. [----------] Global test environment set-up. [----------] 2 tests from GraphTransformationTests [ RUN ] GraphTransformationTests.MatMulIntegerToFloatTest [ OK ] GraphTransformationTests.MatMulIntegerToFloatTest (13 ms) [ RUN ] GraphTransformationTests.MatMulIntegerToFloat16Test [ OK ] GraphTransformationTests.MatMulIntegerToFloat16Test (4 ms) [----------] 2 tests from GraphTransformationTests (20 ms total) [----------] Global test environment tear-down [==========] 2 tests from 1 test suite ran. (22 ms total) [ PASSED ] 2 tests. memleakdbg: ----- No memory leaks detected ----- ``` ### Motivation and Context <!-- - Why is this change required? What problem does it solve? - If it fixes an open issue, please link to the issue here. -->
AnaghaRaoAMD
added a commit
that referenced
this pull request
Nov 22, 2023
MatMulIntegerToFloat is updated to support FP16. The nodes for FP16 Transformation use "Mul" FP16, which is not directly supported by the CPU. For now FP16 transformation is only supported for DML EP. Disabled all FP16 tests on CPU. Tests result without `-use_dml` build flag ``` onnxruntime_test_all.exe --gtest_filter="*MatMulIntegerToFloat*" Note: Google Test filter = *MatMulIntegerToFloat* [==========] Running 8 tests from 4 test suites. [----------] Global test environment set-up. [----------] 1 test from CPU_U8S8_Precision_Tests [ RUN ] CPU_U8S8_Precision_Tests.MatMulIntegerToFloat [ OK ] CPU_U8S8_Precision_Tests.MatMulIntegerToFloat (181 ms) [----------] 1 test from CPU_U8S8_Precision_Tests (181 ms total) [----------] 1 test from GraphTransformationTests [ RUN ] GraphTransformationTests.MatMulIntegerToFloatTest [ OK ] GraphTransformationTests.MatMulIntegerToFloatTest (17 ms) [----------] 1 test from GraphTransformationTests (17 ms total) [----------] 1 test from QDQTransformerTests [ RUN ] QDQTransformerTests.MatMulIntegerToFloat [ OK ] QDQTransformerTests.MatMulIntegerToFloat (656 ms) [----------] 1 test from QDQTransformerTests (656 ms total) [----------] 5 tests from MatMulIntegerToFloat [ RUN ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_U8X8 [ OK ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_U8X8 (195 ms) [ RUN ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_U8X8 [ OK ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_U8X8 (206 ms) [ RUN ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_S8S8 [ OK ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_S8S8 (107 ms) [ RUN ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_S8S8 [ OK ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_S8S8 (114 ms) [ RUN ] MatMulIntegerToFloat.MatMulInteger_With_ZeroPoint [ OK ] MatMulIntegerToFloat.MatMulInteger_With_ZeroPoint (227 ms) [----------] 5 tests from MatMulIntegerToFloat (854 ms total) [----------] Global test environment tear-down [==========] 8 tests from 4 test suites ran. (1713 ms total) [ PASSED ] 8 tests. memleakdbg: ----- No memory leaks detected ----- ``` ``` onnxruntime_test_all.exe --gtest_filter="GraphTransformationTests.MatMulIntegerToFloat*" Note: Google Test filter = GraphTransformationTests.MatMulIntegerToFloat* [==========] Running 2 tests from 1 test suite. [----------] Global test environment set-up. [----------] 2 tests from GraphTransformationTests [ RUN ] GraphTransformationTests.MatMulIntegerToFloatTest [ OK ] GraphTransformationTests.MatMulIntegerToFloatTest (13 ms) [ RUN ] GraphTransformationTests.MatMulIntegerToFloat16Test [ OK ] GraphTransformationTests.MatMulIntegerToFloat16Test (4 ms) [----------] 2 tests from GraphTransformationTests (20 ms total) [----------] Global test environment tear-down [==========] 2 tests from 1 test suite ran. (22 ms total) [ PASSED ] 2 tests. memleakdbg: ----- No memory leaks detected ----- ``` <!-- - Why is this change required? What problem does it solve? - If it fixes an open issue, please link to the issue here. -->
AnaghaRaoAMD
added a commit
that referenced
this pull request
Nov 22, 2023
MatMulIntegerToFloat is updated to support FP16. The nodes for FP16 Transformation use "Mul" FP16, which is not directly supported by the CPU. For now FP16 transformation is only supported for DML EP. Disabled all FP16 tests on CPU. Tests result without `-use_dml` build flag ``` onnxruntime_test_all.exe --gtest_filter="*MatMulIntegerToFloat*" Note: Google Test filter = *MatMulIntegerToFloat* [==========] Running 8 tests from 4 test suites. [----------] Global test environment set-up. [----------] 1 test from CPU_U8S8_Precision_Tests [ RUN ] CPU_U8S8_Precision_Tests.MatMulIntegerToFloat [ OK ] CPU_U8S8_Precision_Tests.MatMulIntegerToFloat (181 ms) [----------] 1 test from CPU_U8S8_Precision_Tests (181 ms total) [----------] 1 test from GraphTransformationTests [ RUN ] GraphTransformationTests.MatMulIntegerToFloatTest [ OK ] GraphTransformationTests.MatMulIntegerToFloatTest (17 ms) [----------] 1 test from GraphTransformationTests (17 ms total) [----------] 1 test from QDQTransformerTests [ RUN ] QDQTransformerTests.MatMulIntegerToFloat [ OK ] QDQTransformerTests.MatMulIntegerToFloat (656 ms) [----------] 1 test from QDQTransformerTests (656 ms total) [----------] 5 tests from MatMulIntegerToFloat [ RUN ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_U8X8 [ OK ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_U8X8 (195 ms) [ RUN ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_U8X8 [ OK ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_U8X8 (206 ms) [ RUN ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_S8S8 [ OK ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_S8S8 (107 ms) [ RUN ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_S8S8 [ OK ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_S8S8 (114 ms) [ RUN ] MatMulIntegerToFloat.MatMulInteger_With_ZeroPoint [ OK ] MatMulIntegerToFloat.MatMulInteger_With_ZeroPoint (227 ms) [----------] 5 tests from MatMulIntegerToFloat (854 ms total) [----------] Global test environment tear-down [==========] 8 tests from 4 test suites ran. (1713 ms total) [ PASSED ] 8 tests. memleakdbg: ----- No memory leaks detected ----- ``` ``` onnxruntime_test_all.exe --gtest_filter="GraphTransformationTests.MatMulIntegerToFloat*" Note: Google Test filter = GraphTransformationTests.MatMulIntegerToFloat* [==========] Running 2 tests from 1 test suite. [----------] Global test environment set-up. [----------] 2 tests from GraphTransformationTests [ RUN ] GraphTransformationTests.MatMulIntegerToFloatTest [ OK ] GraphTransformationTests.MatMulIntegerToFloatTest (13 ms) [ RUN ] GraphTransformationTests.MatMulIntegerToFloat16Test [ OK ] GraphTransformationTests.MatMulIntegerToFloat16Test (4 ms) [----------] 2 tests from GraphTransformationTests (20 ms total) [----------] Global test environment tear-down [==========] 2 tests from 1 test suite ran. (22 ms total) [ PASSED ] 2 tests. memleakdbg: ----- No memory leaks detected ----- ``` <!-- - Why is this change required? What problem does it solve? - If it fixes an open issue, please link to the issue here. -->
smk2007
pushed a commit
that referenced
this pull request
Dec 6, 2023
… (#18554) [Cherry Pick Reviewed] MatMulIntegerToFloat is updated to support FP16. The nodes for FP16 Transformation use "Mul" FP16, which is not directly supported by the CPU. For now FP16 transformation is only supported for DML EP. Disabled all FP16 tests on CPU. Tests result without `-use_dml` build flag ``` onnxruntime_test_all.exe --gtest_filter="*MatMulIntegerToFloat*" Note: Google Test filter = *MatMulIntegerToFloat* [==========] Running 8 tests from 4 test suites. [----------] Global test environment set-up. [----------] 1 test from CPU_U8S8_Precision_Tests [ RUN ] CPU_U8S8_Precision_Tests.MatMulIntegerToFloat [ OK ] CPU_U8S8_Precision_Tests.MatMulIntegerToFloat (181 ms) [----------] 1 test from CPU_U8S8_Precision_Tests (181 ms total) [----------] 1 test from GraphTransformationTests [ RUN ] GraphTransformationTests.MatMulIntegerToFloatTest [ OK ] GraphTransformationTests.MatMulIntegerToFloatTest (17 ms) [----------] 1 test from GraphTransformationTests (17 ms total) [----------] 1 test from QDQTransformerTests [ RUN ] QDQTransformerTests.MatMulIntegerToFloat [ OK ] QDQTransformerTests.MatMulIntegerToFloat (656 ms) [----------] 1 test from QDQTransformerTests (656 ms total) [----------] 5 tests from MatMulIntegerToFloat [ RUN ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_U8X8 [ OK ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_U8X8 (195 ms) [ RUN ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_U8X8 [ OK ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_U8X8 (206 ms) [ RUN ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_S8S8 [ OK ] MatMulIntegerToFloat.HasZeroPoint_NoBias_test_S8S8 (107 ms) [ RUN ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_S8S8 [ OK ] MatMulIntegerToFloat.NoZeroPoint_HasBias_test_S8S8 (114 ms) [ RUN ] MatMulIntegerToFloat.MatMulInteger_With_ZeroPoint [ OK ] MatMulIntegerToFloat.MatMulInteger_With_ZeroPoint (227 ms) [----------] 5 tests from MatMulIntegerToFloat (854 ms total) [----------] Global test environment tear-down [==========] 8 tests from 4 test suites ran. (1713 ms total) [ PASSED ] 8 tests. memleakdbg: ----- No memory leaks detected ----- ``` ``` onnxruntime_test_all.exe --gtest_filter="GraphTransformationTests.MatMulIntegerToFloat*" Note: Google Test filter = GraphTransformationTests.MatMulIntegerToFloat* [==========] Running 2 tests from 1 test suite. [----------] Global test environment set-up. [----------] 2 tests from GraphTransformationTests [ RUN ] GraphTransformationTests.MatMulIntegerToFloatTest [ OK ] GraphTransformationTests.MatMulIntegerToFloatTest (13 ms) [ RUN ] GraphTransformationTests.MatMulIntegerToFloat16Test [ OK ] GraphTransformationTests.MatMulIntegerToFloat16Test (4 ms) [----------] 2 tests from GraphTransformationTests (20 ms total) [----------] Global test environment tear-down [==========] 2 tests from 1 test suite ran. (22 ms total) [ PASSED ] 2 tests. memleakdbg: ----- No memory leaks detected ----- ``` <!-- - Why is this change required? What problem does it solve? - If it fixes an open issue, please link to the issue here. --> ### Description <!-- Describe your changes. --> ### Motivation and Context <!-- - Why is this change required? What problem does it solve? - If it fixes an open issue, please link to the issue here. -->
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
MatMulIntegerToFloat is updated to support FP16. The nodes for FP16 Transformation use "Mul" FP16, which is not directly supported by the CPU.
For now FP16 transformation is only supported for DML EP. Disabled all FP16 tests on CPU.
Tests result without
-use_dmlbuild flagMotivation and Context