CuPy or other parallel methods for GPU usage on TX2?

I’m trying to make a project using the CUDA cores and show parallel speedup however it seems I can’t install any of the tools I’ve seen in other projects including CuPy(Don’t know if unsupported), Anaconda(not supported by Jetson platform?) and other tools. I’m wondering if there’s a way to install Cupy or any good tutorials on parallel GPU computing using the Jetson TX2. I wanted to originally use opencv and the CUDA cores to show how much speedup the Jetson TX2 gives compared to a regular CPU single core. Any advice appreciated.

Hi @Bewlar, if you use the l4t-ml container, it already comes with CuPy installed in the container. The latest versions of this l4t-ml container also include OpenCV that was built with CUDA support.

If you don’t want to use container, you can follow along with the steps I used to install CuPy here:

Also to manually build/install OpenCV with CUDA enabled, you can use @mdegans build scripts:

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.