Isn't this the same CPU/FPGA that is on the Parallella? The 7020? Should provide a decent sized ecosystem if it is.
Early publicity photos of Parallella had this exact dev board in them. :)
Zedboard, dual ARM cores with an FPGA attached
11–15 of 15 posts
Re: Zedboard, dual ARM cores with an FPGA attached
#12Earlier quoted context omitted.
Chuck, have you verified that the price is $15? because i heard it's mostly marketing and the real price is much higher.
Well it was a quote I got from the distributor. At a conference the guy talking about the chip gave me that price and I asked the distributor to verify it. They squirmed a bit but eventually said they would (also Arrow btw). That said, there has apparently been a lot more interest in this chip than Xilinx expected and I got the impression they were worried they had under priced it. But some of the markets they are sh…
I think the real challenge for a kickstarter would be to offer attractive fpga development route. Currently xilinx prices their fpga-c dev tool at around $5K. My guess is that's the real limit , not the price of the zedboard.
Re: Zedboard, dual ARM cores with an FPGA attached
#13So I submit this because I just bought one ($395 from Arrow) and the parts are supposed to be like $15 in quantity. My goal is to build a simple framebuffer in it such that you can have not only an ARM SOC with memory controllers, gigabit ethernet, etc. But also a well documented (and actually changeable) way to display stuff. Prior to finding out about this chip I was sort of half-heartedly working on a simple 1080p…
Interesting. Zynq-7020 is supposedly less than $50 in reasonable quantity, if you stay firm with the distributors. Xilinx are known for unpublished discounts, using the discount code Altera :) BTW Couldn't you make a cheap FPGA into a frame buffer, not use a separate ARM chip?
Couldn't you make a cheap FPGA into a frame buffer, not use a separate ARM chip?
Yes you can, integration isn't as nice though. It's pretty reasonable to build a DDR3 memory controller into a Spartan3 or Cyclone4 and the ARM PMI bus. My sketch has the memory slaved to a DMA control dumping RGB888 into a Silicon Labs or TI HDMI 'phy' at the requisite rate to generate a 1080p/60 display. The thought being that the RAM would be 'display preferred' and processor access would be 'when available'. Not as nice as a direct memory interconnect but easy to add to pretty much anyone's standalone ARM chip. I was shooting for something like the Sun3 color framebuffer on the Sun 3/60 which was pretty easy to talk to an pretty much hard to imply I'm violating Patents since those documents are from the 80's.
I couldn't get it much below $15 though, it is a 3 chip solution (RAM / FPGA / Phy). The Zynq absorbs both the RAM and the FPGA so I'm left with the physical interface. And in the Zynq case I actually get two CPU cores so one could conceivably do a full 3D software implementation in one CPU and have the other one run Linux and talk to it via a simple in memory mailbox protocol.
The Zynq is also already soldered together and trying to get the Digilent plugon modules talking to the Spartan 3 dev board i had was looking dicey at best. :-) I can fantasize about Atmel doing a hard core of a simple Frame buffer but its not realistically in the 'likely' category.
Re: Zedboard, dual ARM cores with an FPGA attached
#14So I submit this because I just bought one ($395 from Arrow) and the parts are supposed to be like $15 in quantity. My goal is to build a simple framebuffer in it such that you can have not only an ARM SOC with memory controllers, gigabit ethernet, etc. But also a well documented (and actually changeable) way to display stuff. Prior to finding out about this chip I was sort of half-heartedly working on a simple 1080p…
I've heard this so many times that I've grown tired of it. The "$15 in quantity" claim from Xilinx and other FPGA manufacturers. Yes, it is quite likely that after 2 years your order of 100,000 units (first batch in a larger 500k schedule, of course) will eventually reach that mythical $15.
I've learned to design things using my local distributor as a pricing source. I use Farnell and look at their qty 10 pricing. That, together with the fact that a part is available at all, is a good indicator: I know I can get the pricing down at least 40% if I achieve quantity. And I have my feet firmly planted on earth.
So, that Zynq chip? Farnell doesn't stock it at all, and findchips shows one source, beginning at $105 per chip. Ouch.
I realize hackers enjoy tinkering with FPGAs, but for most practical applications in this power/performance range you'd be much better off with a Freescale i.MX6, those begin at $15 (quantity 10) and $42 will buy you a quad-core. Those chips have loads of peripherals, pretty much anything you can think of, and pretty impressive graphics accelerators. Which means you end up using pre-designed, tested and documented subsystems instead of developing your own logic.
Re: Zedboard, dual ARM cores with an FPGA attached
#15Earlier quoted context omitted.
Well it was a quote I got from the distributor. At a conference the guy talking about the chip gave me that price and I asked the distributor to verify it. They squirmed a bit but eventually said they would (also Arrow btw). That said, there has apparently been a lot more interest in this chip than Xilinx expected and I got the impression they were worried they had under priced it. But some of the markets they are sh…
I think the real challenge for a kickstarter would be to offer attractive fpga development route. Currently xilinx prices their fpga-c dev tool at around $5K. My guess is that's the real limit , not the price of the zedboard.
Xilinx's official dev boards are all pretty expensive, but they tend to be all-singing all-dancing affairs, with largest-in-class FPGAs, tons of peripherals and interfaces, and enough little blinky lights to blind a goat. Third-party dev boards (like the ones from Digilent: http://digilentinc.com/) tend to be much more reasonable.