I'm a software guy with only the most basic of hardware training/knowledge so while I understand the basic concepts of ASICs and FPGAs the details are beyond me. What are the primary differences and, more importantly to my purposes, how do those differences play out for bitcoin mining purposes?
2 Answers
An FPGA is cheap in quantity one ($100-$300) but expensive in quantity 1,000 ($50,000-$3,000,000). They can mine decently (200MHash/s or so), but are not cost effective yet. (They do, however, consume much less power than GPUs.)
An ASIC is expensive in quantity one ($2,000,000) but cheap in quantity 100,000 ($5,000,000). They can mine unbelievably well (perhaps 1GHash/s), but nobody (as far as we know) has gone to the expense of making one for bitcoin mining yet.
-
There is a company that claims to have an ASIC available on the market in September - but many have dismissed those claims as trolling/fear mongering/market manipulation Commented Aug 31, 2011 at 3:34
-
2I've actually talked to some people about financing such a project. The general response I get is that the numbers are sound but the risk it just too high. They're afraid someone else will come out with an ASIC (increasing difficulty, saturating the market) or the value of Bitcoins will crash before the project can deliver. Commented Aug 31, 2011 at 10:36
-
1What kinda power consumption are we looking at with a single ASIC? Im aware FPGA boards are around 15w for 200Mhash– MaxSanCommented Sep 2, 2011 at 16:25
-
2One would expect ASIC power consumption to be less than a comparable FPGA because they're made with roughly the same technologies but an FPGA will have a lot of stuff you have to keep on but aren't using. Commented Sep 3, 2011 at 1:35
-
3@David Schwartz the upfront doesn't need to be as high as $2M - that would be closer to the cost of a standard cell ASIC. If you use structured ASICs you can get NRE under $1M. In any case, your point is valid that the up front cost is just enormous compared with FPGAs. Commented Sep 16, 2011 at 18:52
I co-founded LargeCoin, Inc., and we really are developing an ASIC for Bitcoin mining. As earlier replies pointed out, the unit cost of ASICs is far lower than FPGAs once you get beyond a few hundred units. The up-front cost of ASIC production varies widely depending on the process you use. On the cheap side, there is so-called "wafer sharing," where your design gets populated on to a wafer with many other designs. Wafer sharing typically uses very old processes that aren't going to deliver the power and density benefits of newer processes. BTW, a "process" refers to a the physical process for making the transistors within the ASIC. Generally a process is referred to by its transistor geometry size - for instance 45nm is a relatively current process.
Next up is "Structured ASICs". A structured ASIC vendor provides an assortment of standard logic components on the bottom layers of the chip. You then add a few metal layers on top of these connecting up the basic components to implement your custom logic. Structured ASICs cost between $100,000 and $500,000 (very approximately) for the first prototypes.
Then there are "standard cell" ASICs, where the chip is designed using standard components that are sort of analogous to software libraries.
And finally, there is "full custom". With a fully custom chip, you are designing everything about the ASIC yourself. The NRE costs for a full custom design is in the millions of dollars typically - particularly if you want to use a modern process.
What process are we using? We're keeping that close to our chest at the moment. However, I can confirm that our detailed modelling at this point indicates we'll be able to mine 250 GHash/s in a single rack of mining units using 5kW of power. When you consider that this represents the computational power of about 400 AMD Radeon 6990 GPUs (which would consume close to 200 kW), you can immediately see the benefit of ASICs for Bitcoin mining.
By the middle of 2012, ASIC mining will be a substantial factor in the Bitcoin economy.
-
4Citation needed for "By the middle of 2012, ASIC mining will be a substantial factor in the Bitcoin economy." Commented Sep 12, 2011 at 16:55
-
2
-
7LoL no, you can't cite prophecy. Look, I understand the skepticism. It's warranted after all the snake oil around ASICs. We'll post hard data when we have it - until then I suppose I'm just a prophet. Commented Sep 14, 2011 at 0:17
-
3
-
3