Scott McKay

@snnn any reason not to use FindPackage for the CUDA libraries? We might need to set CUDAToolkit_ROOT to the cuda_home value when calling cmake though. https://cmake.org/cmake/help/latest/module/FindCUDAToolkit.html

Isn't the 'create' method part of InferenceSessionFactory not InferenceSession? https://github.com/microsoft/onnxruntime/blob/c12a20bef95df5437189687b94e7ba2f1bad1505/js/common/lib/inference-session.ts#L415-L458

Not following what you mean by the 'usage patterns' are the same. I'm not an expert on the JS API but I don't see a `create` method on the InferenceSession...

@fs-eire might be able to help

If you set the default logger severity to VERBOSE what does it say about the unsupported nodes? This needs to be done when creating the environment - i.e. first call...

Changing the data type shouldn't change the rank of any values so the 3D input will still be 3D. The usability checker is intended as a rough guide. It's in...

The 2D MatMul should work with NNAPI without needing to use QLinearMatMul. Was there log output saying why it was considered unsupported immediately prior to the 'unsupported operator' error? u8u8...

There's an issue with how groups of QDQ nodes (2xDQ -> MatMul -> Q) are being processed. We're looking at nodes in their topological order, but this isn't good enough...

AFAIK Qualcomm chips can work with NNAPI but it's up to the chip vendor to implement the low level NNAPI interface on a chip by chip basis. I wouldn't have...

Typically support for specific devices is done by creating an [execution provider](https://github.com/microsoft/onnxruntime/blob/main/include/onnxruntime/core/framework/execution_provider.h). https://onnxruntime.ai/docs/execution-providers/ https://onnxruntime.ai/docs/execution-providers/add-execution-provider.html The GetCapability method implements the logic to tell ORT which nodes you can execute (typically based...