search

deepwavedigital / gr-wavelearner

Perform deep learning inference on signals
GNU General Public License v3.0
59 stars 26 forks source link

Jetson Nano compatibility / port #4

Open vielmetti opened 5 years ago

vielmetti commented 5 years ago

I have a Jetson Nano, hoping to try this out on same - will let you know more details of my installation, and what I find.

I also have the gr-kiwisdr module installed, so I don't even need to have a local radio/antenna to do the work, since there are KiwiSDR's around the world.

vielmetti commented 5 years ago

Noting this:

https://twitter.com/adamlikesai/status/1124039738875437056

fpgasdr commented 3 years ago

Hi, I want to evaluate also this platform. any advise?.

now I tried with gnuradio 3.8.2 but have some incompatibilities(also with gr-cuda). https://github.com/deepwavedigital/gr-cuda/issues/1

dicta commented 3 years ago

@eljuligallego This repository contains software for interfacing TensorRT to GNU Radio for inference. It's expected that users already have some familiarity with TensorRT and have created and trained a network using an (external) machine learning framework. The PLAN files themselves are specific to the particular hardware being used as well as exact version of the TensorRT library in use, so we don't ship them with this repository. Once you have the network trained that you would like to deploy, the following links document the workflow in getting that network ready for deployment as a PLAN file:

If you're curious about how TensorRT fits into the machine learning workflow overall, this page gives a quick summary:

Where does TensorRT fit in?

fpgasdr commented 3 years ago

Thanks @dicta you made a great explanation.

Also sorry for delete the message, I was not sure about it was the wright place to place it. Maybe I can make a new issue to clarify the error. and link your answer.