Are PYNQ concept worth to move onto general FPGA with softcore CPU apart from SOC-based FPGA?

Well it is really good question as truly speaking XRT is the heart of what PYNQ is relying.

However any chance that this Idea can be moved to a ubuntu soft-core FPGA with partial reconfiguration?

This is asked before help post:
PYNQ MicroBlaze

No, No, No PYNQ will not support other than ZYNQ or ZYNQMP we know it.

BUT hold on~

I will try to show what the current situation on two major paths of this idea.

Experiment A - Petalinux + MicroBlaze.
After implemented all the design (ethernet etc) and boot the default Linux. I already know this is a complete dead-end on this path (so a big NO NO).

Experiment B - RISC-V + Ubuntu
After trying out the default RISC-V Xilinx soft-core example (come with Debian) the move over of Ubuntu is truly success and even Jupyter-Notebook is able to run after boot. (So a big YES YES)

With these two experiments we can see the only remain is partial reconfiguration API and how poor is the Jupyter notebook behave.

From this point a single RISC-V big core is truly slow (REAL SLOW), so boot the clock-frequency or increase cores are only the way.

But does this really worth?
So logically-speaking:
Z-7020 ~= XC7A75T but 1 big RISC-V core is already consume over 86% of the XC7A75T
Conclusion no real application value in this point!

Unless you really trying out other CPU architecture and testing out the sanity in that case you are having Virtex on hand and the above path do make sense.

1 Like