I recently obtained the Zynq-7000 AP SoC ZC702 Evaluation Kit, and set it so that I can boot the ‘busybox’ Linux build (part of the Sobel Filter demonstration software provided on the platform).
The experience of seeing the ‘zynq>’ prompt (Linux shell) takes me way back to my days at the University of California at Berkeley, where as a senior in EECS I was on the ‘computer committee’ that invited Ken Thompson of Bell Labs to visit and place Unix on one of the very first PDP-11/70’s. As you may know, 1 MIPs (1 million instructions per second) was a really big deal in those days, and the PDP-11/70 met that performance goal (1 microsecond per instruction – Wow!).
To now have TWO ARM Cortex-A9 cores with their floating point units running Symetric Multiprocessing (SMP) Linux at 800 MHz, sitting here on my desk, consuming less than one watt, almost seems like a miracle when I think back to those first heady days.
Putting it All Together
I have also been installing the various software environments so I can build my own version of Linux, as well as build my own hardware + software “bare metal” (no operating system) designs. I encourage everyone who is working with the ZC702 board, or the new ZED board from Avnet, to post their successes on the forums, and reveal how they overcame their obstacles.
Getting it all to work is somewhat of a challenge, as one has to navigate not only the software land of the Linux build process, but also the new hardware land of the processor system (PS) on the Zynq device. Thankfully, there are demos, wizards, and graphical user interfaces to help you get it all working.
In the Zynq device, the PS is separated from the FPGA portion, which is known as Programmable Logic (PL). Thus, you can boot your software, open the console, and connect via a serial port or Ethernet, all without doing anything at all with the PL side. Once everything is running, you may then configure the PL from the processor (PS). Working in this way may be foreign to FPGA device users of the past, but I for one will never go back (to not having an operating system).
The Adventure Begins
I am sure there are many out there who have a whole system on a chip (SoC) planned for their Zynq device in their system. I have spoken to those who are using the Zynq AP SoC for a variety of applications. Use of the ZC702 board has allowed them to define their software architecture and begin software development before their custom Zynq board has even arrived. Once their custom board is back, they have had remarkably successful experiences with having their applications boot and run.
If they had decided to fabricate their own SoC, they might have had to spend years just to get their silicon, and then a few more years to get their systems built, programmed, and debugged, not to mention the 100M+ dollars it would take to do the job at 28 nm.
Back to the Past
“Microprocessor Interfacing Techniques” by two obscure engineers, Austin Lesea and Rodnay Zaks, appeared in print for the first time in 1977. It was the book everyone was waiting for; the microprocessor had hit the scene and everyone needed a practical book on how to use them. Those were exciting times. It was standing room only at the West Coast Computer Faire in San Jose just to hear me talk about how to do something useful with the new chips from Intel, Motorola, and other microprocessor vendors.
So now, the race is on. All of those applications that were just crying out for an SoC, but you could not afford to make one for, may now be turned into products -- from automotive driver assistance, to solar array controllers. I do not know when I have been more excited about a technology, 1977 or 2012.