Does 'TensorRT 3 Release Candidate for Jetson Platforms' support tensorflow?

Autonomous Machines Jetson & Embedded Systems Jetson TX2
#1

Hi all,

Does ‘TensorRT 3 Release Candidate for Jetson Platforms’ support tensorflow now?, I can not find python api in it.

#2

Hi,

Currently, Python API is only available on the x86-based Linux system.

The recommended flow to run TensorFlow model on Jetson is:

  1. Export Tensorflow model to UFF on an x86 machine with python interface.
  2. Read UFF model into TensorRT and create the inference engine on Jetson with C++ interface.

Thanks.

#3

Thank you AastaLL,
Can I use a x86 machine with geforce 1060 gpu to export tensorflow model?

#4

When can we expect Python API on systems like Jetson (10 days, month, 6 months)?

#5

Hi a961532,

TensorRT can run on GTX-1060.
Please install the tensorRT x86 package on the desktop.
Thanks and sorry for the inconvenience.

#6

Hi AastaLLL

Is ‘TensorRT 3 Release Candidate for Tesla GPUs (Sept 25, 2017)’ the tensorRT x86 package?
It seems that it only support Tesla GPUs (K80, P4, P40, P100, and V100), how could it run on Geforce gpu?

#7

Officially, TensorRT x86 package only supports Tesla GPU.

But we have checked this model conversion feature on GeForce GPU, and it works correctly.
You can install the package with GeForce GPU and covert your model into UFF format.

Thanks.

#8