@williambanks so you’re saying that with ZEC if you manage to solve the mem access bottleneck you can use a single chunk of memory to serve many “computing units”?
EDIT: I dont think so, probably because its unsolvable:)
I’m saying that if you can make a GPU miner for a coin, then you can also make an FPGA for that same coin. If you can make an FPGA then there is a direct path to ASIC.
It’s not a matter of what you CAN do, it’s a matter of what you can afford to do and if the cost to do it could be recouped or not.
$10k for the devkit. $50k min for software development. $150k to get 1000 boards into production plus another $500k for the chips themselves. Time frame of 9 months from concept to customer door.
So you’re looking at a minimum cost of $750k and you should probably double that number to $1.5M
That gets you your first 1,000 miners. Your cost is about $8k per unit at that volume. But after that happens then tooling costs are already sunk so your next 1,000 miners are about half that cost or $4k each also the turn around time on batch 2 would be more like 30 - 60 days.
The third batch could get the price down another 50% etc.
Point is you need to know what you’re doing in manufacturing in order to pull this off and you need the financial resources.
Resources that at least for zcash and most other coins would probably never be recouped.
Meantime an FPGA could be expected to do double or triple what a GPU could do. This is assuming you either have a huge 150MB on chip cache or, or your FPGA supports infiniband memory or something.
So assume $1,000 for a high end GPU miner would be the same as $3,000 for an FPGA / ASIC.
You aren’t getting FPGAs at a cost effective price this generation. The only other reason to go FPGA or ASIC then would be if the power savings would be worth it. You’re talking 80 to 100 watts for triple the performance of a GPU. Not sure that’s really worth it.
2 Likes
I hope someone will develop a 3d printer capable of printing things like usb asic miners and the like all in one piece.