ModalAI Forum

@dario-pisanti Hi Dario,

Thanks for sharing your issue. I'm also working with the Qualcomm Neural Processing SDK on VOXL2 Mini.

The error you mentioned could be due to:

Model compatibility issues during conversion.
SDK configuration, especially for GPU.
Double-check the GPU-specific settings in $SNPE_ROOT/bin/envsetup.sh and try running a simpler model to verify the GPU setup.

Have you managed to resolve this, or do you have any additional details? I'm also trying to use the QNN SDK with ONNX runtime on VOXL2 Mini.

Best,
Manu

@Eric-Katzfey I am not able to run pythorch/Onnx model on GPU or NPU, do we have any examples for that?
I could only find the tflite examples.

I was able to run the Light Glue model on docker with CPU.

Hello Thomas,

I am currently working on a project to convert the LightGlue model from PyTorch to ONNX/TensorFlow and optimize it for Qualcomm's DSP on the VOXL 2 Mini. The goal is to achieve efficient model execution with a target performance of 1 FPS, I want to use the docker on the chip.

Can you please guide me on this?

Thank you!

Best,
Manu