[QNN EP] Add ONNX ScatterElements support

[quic-tirupath]

Description

• Translate ONNX ScatterElements as QNN's ScatterElements Op
• Handle unsupported reduction value i.e., "min"
• Add unit tests to verify ScatterElements Op support on HTP

[HectorSVC]

/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline,Windows x64 QNN CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 5 pipeline(s).

[HectorSVC]

Please check the doc to make sure there are no steps missing: https://onnxruntime.ai/docs/execution-providers/QNN-ExecutionProvider.html#add-new-operator-support-in-qnn-ep
For example, make sure the DQ->CumSum->Q is treat as a single node unit, so it's running as quantized CumSum instead of FP32 in FP16 mode on HTP. Another recent example is: #24659

[quic-tirupath]

Please check the doc to make sure there are no steps missing: https://onnxruntime.ai/docs/execution-providers/QNN-ExecutionProvider.html#add-new-operator-support-in-qnn-ep For example, make sure the DQ->CumSum->Q is treat as a single node unit, so it's running as quantized CumSum instead of FP32 in FP16 mode on HTP. Another recent example is: #24659

Thanks for pointing the missing functionality. The PR is updated with adding missing functionality.

[HectorSVC]

/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline,Windows x64 QNN CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 5 pipeline(s).

[Copilot]

Pull Request Overview

This PR adds support for the ONNX ScatterElements operation to the QNN execution provider. It includes translation of the operator to QNN's ScatterElements op, handling of unsupported reduction modes (specifically "min"), and comprehensive test coverage.

Key changes:

• Implements QNN EP support for ScatterElements with proper attribute handling (axis and reduction)
• Adds validation to reject unsupported "min" reduction mode
• Extends test infrastructure to support multi-type input operations and adds comprehensive unit tests

Reviewed Changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
onnxruntime/test/providers/qnn/simple_op_htp_test.cc	Adds comprehensive test cases for ScatterElements on CPU and HTP backends
onnxruntime/test/providers/qnn/qnn_test_utils.h	Extends test infrastructure with new template functions for multi-type input operations
onnxruntime/test/providers/cpu/tensor/scatter_op_test.cc	Updates existing tests to exclude QNN EP for unsupported scenarios
onnxruntime/core/providers/qnn/builder/opbuilder/simple_op_builder.cc	Implements ScatterElements operator translation with attribute processing
onnxruntime/core/providers/qnn/builder/opbuilder/base_op_builder.h	Adds ONNX to QNN operator type mapping
onnxruntime/core/providers/qnn/builder/op_builder_factory.cc	Registers ScatterElements operator builder
onnxruntime/core/optimizer/qdq_transformer/selectors_actions/shared/utils.cc	Adds QDQ selector registration for ScatterElements
onnxruntime/core/optimizer/qdq_transformer/selectors_actions/qdq_selectors.h	Declares ScatterElements QDQ selector class
onnxruntime/core/optimizer/qdq_transformer/selectors_actions/qdq_selectors.cc	Implements ScatterElements QDQ validation logic

Comments suppressed due to low confidence (1)

onnxruntime/test/providers/qnn/simple_op_htp_test.cc:1363

• Test name should follow consistent capitalization. Change 'int8' to 'Int8' to match the naming pattern used in other tests like 'ScatterElements_Int8_Reduction_Max'.

TEST_F(QnnHTPBackendTests, ScatterElements_int8_reduction_mul) {

[HectorSVC]

Need to disable some Onnx node test: scatter_elements_with_negative_indices

[quic-tirupath]

Need to disable some Onnx node test: scatter_elements_with_negative_indices

We are checking how can we disable it.
Adding below entry in file:onnx_backend_test_series.py is not helping.
"current_failing_tests_QNN": [ "^test_scatter_elements_with_negative_indices" ],

[quic-tirupath]

Need to disable some Onnx node test: scatter_elements_with_negative_indices

We are checking how can we disable it. Adding below entry in file:onnx_backend_test_series.py is not helping. "current_failing_tests_QNN": [ "^test_scatter_elements_with_negative_indices" ],

@HectorSVC
this is fixed now. Could you please help to retrigger the CI job

[HectorSVC]

/azp run Linux QNN CI Pipeline,Win_TRT_Minimal_CUDA_Test_CI,Windows ARM64 QNN CI Pipeline,Windows GPU Doc Gen CI Pipeline,Windows x64 QNN CI Pipeline

[azure-pipelines]

Azure Pipelines successfully started running 5 pipeline(s).