/electronics/ - Electronics engineering

I currently own a spartan 6 lx25 for general HDL practice and interesting high speed projects.

I have worked with a Nexys3 board in the university, I'm about to try implement a crypto coprocessor in a XUP-V5.
I also got a ZYBO for myself, it should arrive soon

>>8
ZYBO with the xillinx zync? I would love to get one of those when I can afford it. Just the thought of mixing the specialized processing power of an fpga with a general purpose architecture is very intriguing. Imagine being able to upgrade your processor architecture with a simple firmware update.

>>21
>Altera
>qeq

But it looks like a solid dev board.

Ive never messed with any of this stuff before.
What are they generally used for?
What can you do with them that you cant with an mcu?

>>50
FPGA's are needed in all projects where timing is very tight. To explain them, I'll assume you know how microchips are just a huge network of semiconductors. FPGA's are chips with no logic on them to begin with, it has to be programmed in by the user.

On a micro controller you write a program in software that runs on its hardware, on an FPGA you write the hardware.

Why must all FPGAs require nonfree software to program?
Why don't they just take some flash and stick it in the same package so you don't need a separate chip?
While they're at it, why don't they stick a microcontroller on there so people don't use half their gates on a nonprogrammable CPU?
They're basically useless.

>>105
you need silicon level knowledge to write software for FPGA's. Why would any manufacturer release such detailed info? They would be copied to shit. I'm pretty sure you can buy FPGA's with built in micros, but why would you when it is much cheaper to buy a mass produced micro for less than a dollar?

>>106
>you need silicon level knowledge to write software for FPGA's. Why would any manufacturer release such detailed info?
This sort of reasoning is the reason there's so much NDA shenanigans around, and it's almost surely bullshit. Having a few specs is not enough for anybody to just go out and start making their own. I expect most of those "super secret optimizations so advanced we can't tell you what they are" are 100% placebo.
>why would you when it is much cheaper to buy a mass produced micro for less than a dollar?
Sticking a few arm cores on the side costs basically nothing for the chip manufacturer. They're tiny. The microcontrollers may be cheap but board space and wiring is not free. I guess maybe the FPGAs have some process differences that make it harder, but that sounds like a copout.

But then FPGAs seem to be overpriced in about every way possible, which is unfortunate because it seems like there would be so many uses for them.

>>107
>Sticking a few arm cores on the side costs basically nothing for the chip manufacturer.
Not for licensing from ARM.
>I guess maybe the FPGAs have some process differences that make it harder, but that sounds like a copout.
FPGAs are single process devices almost all of the time, they're in the range of 22nm where as MCUs from your favorite vendor are 90nm+, the number of passes to have a fully functional microprocessor and peripherals would make it as expensive as buying both(take a look at any currently available hybrids) and how a much higher yield rate.
>But then FPGAs seem to be overpriced in about every way possible
not really, you're getting 2-8x the die for something completely incomparable to an mcu, the silicon cost is the same as any CMOS process.