Take the PYNQ Poll: You know Python, but Do You Know PYNQ?

Greetings!

While the vast majority of my programming has been in C, I have used C++ on some projects.  In one my most memorable projects, I was following the lead of my manager, who was an absolute master in the concepts of reuse, where he had over the years constructed (and reused on several projects) a framework that allowed remote control of the device, which proved to me invaluable.  We could peek and poke into memory, call function/procedures, load functions into memory and call them, etc.  He constructed this framework in C++.  My main part of the project was image processing, which used a lot of C++ function template for sorting and the such, that made the coding much easier (it also used a lot of memory as there was a lot of overhead to these functions).

Gene

genebren Hi!

You mentioned that you are content with C++ and so I have to ask. What is your application area for the language?

I am asking from the work point of view. Though I am an electronics designer who if proficient at a bunch of languages, I am looking for the correct way to 'resume' my skills.

Thanks in advance.

Pynq is something I need to learn. The syntax is python like which is good. The unique features are embedded in the libraries. For me that is a doable learning curve. So into the bucket list.

Thanks, James. That makes it clearer - and makes a lot more sense. I just watched your video on the road test too.

[Python] does seen a strange match for FPGA developmentIt does seem like a strange match up at first, but not so much when you see how it works together. Python isn't used for developing the FPGA, or what the Zynq calls the "programmable logic" layer. IP is still developed using the Xilinx Vivado toolchain.

Python is used to load overlays which contain the FPGA/PL IP. It then acts as the glue between the hard ARM-A9 core and the FPGA core.

In that context, Python makes much more sense. FPGA blocks are developed in VHDL/Verilog/RTL/HLS/etc. Code on the application processor, with the PYNQ image, runs Python under linux. The tight bindings accelerates gluing a hardware FPGA operation to a software function. The main example I looked at was OpenCV. The FPGA/PL layer communicates with HDMI and USB passing data over to the ARM processor. There, some Python code uses OpenCV to do facial recognition. It takes about 20 lines of Python code to make it work.

For zynq(unless we are using just the PS), C hardly can be used. It has support via the old HLS and currently SDSOC. PYNQ's Python is somewhat matching SDSOC in simplicity in my opinion but for a maker PYNQ is more welcoming. My client who try to use zynq for image processing have spent up to a year just to accomplish the image pipeline with VHDL and hardware prototype while the PYNQ can be used side by side for the algorithmn development to be carried out concurrently while waiting for the hardware prototype to be ready.

But now there is a major and cheaper contender in the form of NVIDIA Jetson Nano at $99

i use c, c++, c# much more than Python. But nowadays many things are good to start with Python. Pynq is one. However, i prefer to use distribution like Anaconda to deal with Python

Answering my own question - it looks like your can run PYNQ on any Zynq board, but there will be some effort required building an SD card image. Only the Pynq Z1 and Z2 boards have prebuilt images available.

I'm not a huge fan of Python and it does seen a strange match for FPGA development. I'll be watching the webinsr with interest to see if I can be convinced.

The hardware looks nice, but I assume that the PYNQ framework will run on any Zync board such as the MiniZed

That should probably say ...

"Other; please explain in comments (and I can't wait to attend the webinar)"

I'd have to agree that I'm happy with C myself, but I can see that Python is quite popular and has a lot of conveniences (as well as subtle pitfalls as well, which I'm learning about as I occasionally fall into them). I think it's a good thing that more languages are given the chance to be intimate with hardware - there have been various microcontroller boards which can be programmed in Lua, dotNet, VB, JavaScript, etc. I guess it's a good thing that lets prototype creators use what they are familiar with so that they don't (necessarily) have to learn an entirely different language ... many students nowadays are not necessarily starting with C anymore even though I think they probably should ...

- Gough