Ken Arnold writes: Yes, I agree that it is a bad idea to use any tool that is a bad impedance match for the problem at hand, whether it's a chip, OS, or a computer language. So, are you volunteering to assist us with a new board with an FPGA from another vendor? ;) It's a sensitive topic and we had quite a bit of internal discussion about this choice, hence my long response below. We will be using other FPGA vendors parts when we can, but we have much still to do and very limited resources. But Money Talks. So far the ONLY contributions have been from the university, me personally, my product development business, and a little bit of cash generated from HiTechEdVentures.org's consulting and crowd-funding activities (so, not very much). And Max was kind enough to inviite us to blog on APP, also sponsored by Xilinx. Total contribution from all FPGA vendors to date = $0.00 and 0 assistance of any other kind beyond what Max has done and vendor reps aand sales people occasionally chatting with us, typically when they needed to log an "account visit" for their bosses in sales management. Xilinx has helped our programs in a very tiny way through their university program and a few samples now and then, which is vastly more help than I can say we received from all the other programmable logic vendors combined, so naturally they were first on our list. That said, one of the things I personally try to instill in all the students is a healthy sense of skepticism, which has apparently been effective. As my friend and former student Randy Hern stated so well: "Data sheets are marketing documents, NOT engineering documents." As much as we can, we try to use technology in our program that is not tied to a specific vendor, such as ARM ISA and open source material from sources like OpenCores.org. And we cover both VHDL and Verilog, but there isn't any vendor agnostic FPGA architecture, so we had to pick something. And while they may publicly claim otherwise, most semi vendors look at university and small enterprise consulting activities with the same short term "they're never going to amount to significant sales (and therefore a sales commission), so it's not worth our effort" point of view, even though some pretty big production runs, new businesses, and whole market segments have grown out of university incubator programs. A few of the semi vendors will occasionally make a big deal over a short term university program, but they often spend more on banners and promoting their efforts than they do actually helping universities. Many vendors have been very supportive of university programs in the past, but that has diminished greatly since they sense that the designs are mostly offshore. So the minimal assistance can dry up very quickly. That's also one of the reasons why we want to spin our own board solutions: Freescale dropped a microcontroller product that one of our faculty was depending on using for his course, and they never even returned his calls and emails asking where we could obtain the eval boards for the course until it was too late. He had to rewrite a massive amount of course and lab material in very little time, which is why we switched to ARM architecture for that course. Any given vendor may choose to "deprecate" a board or chip without warning, but the ARM architecture will persist for a long time. Not so for FPGAs.
While just one vendor is good for the basic's, students also need to be made aware of the pro's and con's of each vendor's technology and the full range of offerings in the FPGA world -- this way one can chose the correct part for the job -- there is nothing like having to tell management you need to add 10,000 pounds of structural strength to a product because the Xilinx FPGA takes 0.25 sec to restart, and a consultant is asking why you did not use another vendor!
Interesting! Such a competition would certainly present the students with a challenge. We'll take a closer look at that, and see what they come up with. One of the best things about the students is that they don't know the "standard and customary" ways to solve problems, so they often surprise me with the "outside the box" and totally unexpected solutions they come up with. Often when I see something they've done, I have to suppress my "that won't work" reflex, and find they have a truly unique solution. That's a big kick for me, and is right up there with seeing the light in their eyes when a difficult concept dawns on them. I learn a lot from our students, and get to have fun too!
Some of them have gone on to design new products, start new companies, and one of them even created a new product category with his class project a while back.
Yes, exactly! -- I hope you're correct -- that's the intent of our program. Knowledge of the theory is necessary, but not sufficient. It's easier to teach the theory alone, because that doesn't change as fast as the technology. But to be effective, engineers must be able to apply the theory and understand why their designs may work in simulation and fail in a real chip.
Besides, making things blink, click, buzz and whir is a whole lot more fun!
Thanks for your interest. We've used other vendors, but Xilinx is more popular in industry, and we want to make sure our students have experience with the dominant suppliers as well as the secondary ones.
We have a university license for Chipscope so the students use that for their projects. Also, we don't like using the suppliers eval boards because they can drop them the moment they redirect their marketing efforts to another part without warning, leaving us holding the bag!
I'd be happy to trade experiences teaching classes, and you can email me ken.eet at hte dot com.
Hi, It is very important that engineering students are able to get there hands on real FPGA boards enabling them to test and see their design working in the silicon. It is very important especially for the times when the FPGA design does not quite function as desired, it then allows the student to learn the important debugging skills and gain confidence in their skills. This programme should help students acheive that goal.
I'm a design engineer with a strong background in FPGA development. Last Fall, I taught a course in digital design (adjunct) and wanted to incorporate an FPGA development kit that the students could keep after the course was finished so they could continue exploring on their own. I wholeheartedly support the inclusion of FPGAs in college courses and love when students express interest in the technology. I think it's fantastic that you can engage so many students with different aspects of this project.
When evaluating Xilinx and Altera, it appeared that Chipscope was not provided with the free version of ISE while SignalTap is part of Quartus Web Edition. This was one of the primary factors that swayed me to use the Altera-based DE0-Nano. The students found SignalTap to be quite effective in helping them troubleshoot their designs. I did start with functional simulation but I wanted the students to encounter some of the differences between simulation and synthesis. Were you able to find a workaround for this?
Blog Doing Math in FPGAs Tom Burke 16 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...