PYNQ Now Available for the Kria KV260 Vision AI Starter Kit

We are happy to announce PYNQ support for the Kria SoM and its official Ubuntu SD Card image.

Starting from that SD Card, run a single install.sh script from the new Kria-PYNQ GitHub repository to get PYNQ’s Pythonic APIs to programmable logic, 4 included hardware overlays and 50+ Jupyter notebooks. Those overlays include our the newly released Composable Pipeline and the ever popular DPU-PYNQ Python package.

If you have a SoM, please give PYNQ a try - you’ll find an identical Python virtual environment installed on the SoM that we install on all other PYNQ v2.7 enabled platforms. Its our hope that this uniform Python environment will allow community members to map existing overlays to the SoM.

And finally, If you have any questions, of course please post on the forum here to access the growing PYNQ community.

Best,
The PYNQ Team

9 Likes

Thanks for sharing, was waiting for the PYNQ release. Excited to try out :smiley:

Hi, excited to try this out - but I’m having problems with the installation and unmet dependencies.

I’m using the Ubuntu 20.04 LTS image.

I have run the snap environment previously and the Xilinx version of Gstreamer. Is that a problem?
Do I need to start from a fresh image?

Ralph

Hi @ralphjy,

I ran into the same issue and started over with a fresh SD card image (but this time without installing the Gstreamer libraries) and I was able to get PYNQ working following your suggestions.

Basically, write the Ubuntu SD card image, boot the board, change your password, verify that networking is up and do not complete the rest of the Booting your Starter Kit section:

Also, here is a link back to the E14 announcement from javagoza where I saw this come across my email feed:

So far it is looking really great! Thank you everyone from the PYNQ team that was involved in making this happen!

-Kevin

1 Like

Hi Kevin,

Just got back from walking the dog and saw your reply. I had also done a fresh install, but just got a chance to try it.

It works for me also, which is great (really like PYNQ). On to trying the example notebooks.

I’ll just keep a separate SD card for the Snap Apps.

I’m having a terrible time getting RTSP inputs working with the GStreamer pipelines on the Kria. I mentioned it to Mario. I had worked with Jason Moss on a similar issue on the UltraZed-EV, but he left Avnet before it got resolved. I think I just don’t know how to configure the pipeline for the VCU or my network is doing something bad.

Kudos to the PYNQ team. Great job!

Ralph

1 Like

Thanks for trying it out - we were verifying on clean sdcards and we’ll update the README and/or install.sh to better coexist with the gstreamer libraries, in particular libdrm-dev…

Thanks a lot for the effort of getting Pynq on Kria.
It’s running. :slight_smile:

1 Like

We are now using the libdrm-xlnx package (commit link)

Please raise issues in new forum posts as they come up!

1 Like