saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-07-24 19:43:35 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Fix Reduced Op build with empty FP16 kernel function tables (#21038)

Browse source 
### Description
- Fixes compilation error for "reduced operator" builds with no FP16
kernels and `MLAS_F16VEC_INTRINSICS_SUPPORTED` enabled.
- Fixes linker error for "reduced operator" builds with QNN EP by
excluding QNN EP unit tests. QNN EP unit tests require CPU EP operator
implementations to evaluate accuracy.


### Motivation and Context
Need to be able to build a reduced operator build with QNN EP. See
https://github.com/microsoft/onnxruntime/blob/main/docs/Reduced_Operator_Kernel_build.md

The following example operator config file causes a compilation error
when either `MLAS_F16VEC_INTRINSICS_SUPPORTED` is defined or QNN EP is
enabled.
```
# reduced_op_config.txt
ai.onnx;12;Add
```

```shell
python tools\ci_build\build.py --include_ops_by_config reduced_op_config.txt --config Debug --build_wheel --build_shared_lib --skip_tests --build_dir build --parallel --use_qnn --qnn_home '<QNN_ROOT_DIR>'
```
This commit is contained in:
Adrian Lizarraga 2024-06-14 14:23:12 -07:00 committed by GitHub
parent f4b22f89bc
commit 8f0e896c95
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 5 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
cmake/onnxruntime_unittests.cmake
View file

@ -654,7 +654,9 @@ if(onnxruntime_USE_JSEP)
list(APPEND onnxruntime_test_providers_libs onnxruntime_providers_js)
endif()
if(onnxruntime_USE_QNN)
# QNN EP tests require CPU EP op implementations for accuracy evaluation, so disable on minimal
# or reduced op builds.
if(onnxruntime_USE_QNN AND NOT onnxruntime_MINIMAL_BUILD AND NOT onnxruntime_REDUCED_OPS_BUILD)
list(APPEND onnxruntime_test_framework_src_patterns ${TEST_SRC_DIR}/providers/qnn/*)
list(APPEND onnxruntime_test_framework_libs onnxruntime_providers_qnn)
list(APPEND onnxruntime_test_providers_dependencies onnxruntime_providers_qnn)

1
onnxruntime/contrib_ops/cpu/cpu_contrib_kernels.cc
View file

@ -178,6 +178,7 @@ Status RegisterNchwcKernels(KernelRegistry& kernel_registry) {
#ifdef MLAS_F16VEC_INTRINSICS_SUPPORTED
Status RegisterFp16Kernels(KernelRegistry& kernel_registry) {
static const BuildKernelCreateInfoFn function_table[] = {
BuildKernelCreateInfo<void>, // default entry to avoid the list become empty after ops-reducing
BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kMSDomain, 1, MLFloat16, FusedConv)>,
BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kMSDomain, 1, MLFloat16, NhwcFusedConv)>,
BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kMSInternalNHWCDomain, 12, MLFloat16, MaxPool)>,

1
onnxruntime/core/providers/cpu/cpu_execution_provider.cc
View file

@ -2711,6 +2711,7 @@ Status RegisterOnnxOperatorKernels(KernelRegistry& kernel_registry) {
#ifdef MLAS_F16VEC_INTRINSICS_SUPPORTED
Status RegisterFp16Kernels(KernelRegistry& kernel_registry) {
static const BuildKernelCreateInfoFn function_table[] = {
BuildKernelCreateInfo<void>, // default entry to avoid the list become empty after ops-reducing
BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 1, MLFloat16,
GlobalAveragePool)>,
BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 11, MLFloat16,