tensorrt-3.0.1-cp27 is not a supported wheel on this platform.

Hi,

I try to get the tensorrt python support in the tx1. But when I run the command ‘pip install tensorrt-3.0.1-cp27-cp27mu-linux_x86_64.whl’, there is an error:
tensorrt-3.0.1-cp27-cp27mu-linux_x86_64.whl is not a supported wheel on this platform.

What does that mean? Is the wheel not supported on tx1?.

Thanks.

Hi,

TensorRT Python API is only available on the x86-based machine, not for Jetson platform.
Sorry for any inconvenience.

Hi,

Whether it is possible to accelerate tensorflow model through UFF on Jetson tx1 using c++? And if so, are there some sample code as tutorial?

Thanks.

Hi,

We have a native sample for exporting a UFF model and creating a TensorRT engine with C++.
Please check: /usr/src/tensorrt/samples$ vim sampleUffMNIST/sampleUffMNIST.cpp

Thanks.

Hi,

I have read the sample. It shows how to load a UFF model with C++. But the first problem is how to conver a tensorflow model to a UFF model with C++, just like ‘uff.from_tensorflow_frozen_model(frozen_file, output_nodes, **kwargs)’ in python. Are there some C++ sample codes showing how to use the conversion tools?

Thanks.

Hi,

You need Python API to convert Tensorflow model into UFF format. It can’t be done via c++ interface.
Please convert Tensorflow model to UFF on the x86 machine first and exporting UFF model on Jetson with the approaches mentioned here.

You can find an example for converting TensorFlow mode to UFF here:
/usr/local/lib/python2.7/dist-packages/tensorrt/examples/tf_to_trt/tf_to_trt.py

Thanks