I'm considering making an FPGA based bitcoin miner - using 28 nm FPGA technology (Altera's Cyclone V).
I believe that this should be able to have a sustained rate of about 1000-1200 MHash per second
(2 FPGA's per card - 301K logic cells each FPGA - running between 250-300 MHz, using a fully unrolled,
fully pipelined SHA256 cores).
Here's my web site :http://www.raspberrycoins.com
I understand that ASIC miners have been announced, and these promise much faster performance - in the
future - when they ship, and apparently companies like Butterfly labs have been taking pre-orders, and
ship your product two months after they take your money. I also don't know if anyone has received
anything except the lowest performance product.
If a ASIC miners are available, then does it doesn't make sense to get an FPGA, but if they're just
pie-in-the-sky (ie: not yet shipping), it might make sense to start doing some mining now, before
ASIC miners come out (and the network hash rate goes up). The first guys who get these ASIC based
miners and start mining before the difficulty level increases significantly have the most to gain,
so a two month wait can be a substantial hit.
We've included a quick ROI calculator at :http://www.raspberrycoins.com/calc
The idea is for an FPGA card with 2 Cyclone V (28 nm FPGA's - 602K logic cells total), connected to
a Raspberry Pi (single board linux computer with ethernet). The solution is scalable, as a single
Raspberry Pi could easily control 4 FPGA cards (8-FPGA's total), and the Raspberry Pi's can be connected
together on a network (using a inexpensive 10/100 base-T switches/hubs). The Raspberry Pi's should be
pretty energy efficient (as they will not be running graphics, but just checking the status of FPGA's).
What do you think? Any feedback and comments are appreciated.