Pandas installation fails on PYNQ Z2

  • PYNQ version: 2.5 or 2.7 (images from the pynq website burnt onto the default PYNQ Z2 sdcard)
  • Board version: PYNQ Z2

I wanted to try the FINN examples on my PYNQ board. When running their install command: pip3 install finn-examples -v we can see that pandas is getting installed.
Unfortunately during the Building wheels for pandas the system throws an error and the entire FINN examples installation fails. More details can be found here.
Installing pandas is however more of a general PYNQ issue then a FINN issue so I was wondering:
How can pandas be successfully installed on the PYNQ Z2 board?

Hi,

If I didn’t recall faultlessly I did tried to install tensorflow (pandsa as well) on PYNQ 2.7 ZYNQ environment and turns out this is a complete dead water as latest tensorflow only up to Python 3.7.
So meantime if you need a Pandas on Python 3.8 with ARM v7l whl to install correctly.

@briansune
Thank you for the manual wheel installation advice.
When using the wheels here the installation completes quickly.
The only issue is that when running pip3 install finn-examples -v it ignores all my previous installed pandas, numpy and pynq packages and reinstalls them by building from source. Is there a way to prevent this?

Ok so I did find out that the actual culprit is the pynq installer that finn-examples invokes.
So just try pip3 install pynq and it leads to the same issue.

Unfortunately I didn’t manage to tell the pynq installer to not reinstall everything.
But I did manage to tell it to use the precompiled wheels for numpy, pandas and cffi.
The steps I took are the following:

  1. Download numpy-1.23.0-cp38-cp38-linux_armv7l.whl, pandas-1.4.3-cp38-cp38-linux_armv7l.whl and cffi-1.15.0-cp38-cp38-linux_armv7l.whl from here.
  2. Run pip3 install --find-links ??? git+https://github.com/Xilinx/finn-examples.git@dev -v where ??? is the path where you downloaded the whl files to. This will make the installer use your precompiled files instead of trying to build everything.

There should however be a better solution to all of these issues.

1 Like

Good to hear that.
So it is confirmed that ARMv7l pandas python 3.8 is good.