Open kchawla-pi opened 3 years ago
which build you are refering to: onnx, tensorflow, deepspeech?
I apologise for the lack of details. I'll correct that now.
I wish to build an onnxruntime python wheel running on a tensorflow backend. I built one for Python3.7 using a virtual environment of Py3.7, but creates a segmentation fault when I import it.
This is on the Xavier AGX (which comes installed with 3.6), other versions are same as yours as well. I don't know if this is because the TensorRT python bindings that come preinstalled with the Jetson are for Py3.6 and I am using a virtual environment using Python3.7.
I tried using the wheel you built: https://github.com/domcross/Jetson-Xavier-AGX-stuff/releases/download/onnx/onnxruntime_gpu_tensorrt-1.0.0-cp36-cp36m-linux_aarch64.whl
And it is importing without error and communicating with the GPU, but it is for Python3.6 and I need it for Py3.7
Thanks you!
I would have tried to build it with a Py3.7-venv as well. Yes, the TensorRT bindings might be the problem here. My current projects do not need onnx/runtime, sorry that I can't help you with that...
Hi, sorry I didn't know how else to ask for your help. Would you share the commands you used to build this? I am trying to build this for use in Python3.7 and mine doesn't work, though it compiled fine.