Also it does NOT take 6+ months to design an ASIC that does something as basic as SHA. Cointerra just started their design a week ago and are going to deliver in December.
you sound quite naive. while your intentions may be honourable, you're living in some dream world where commercial and financial reality doesnt exist.
the goal of an asic company is not just to sell enough chips to pay for its NRE. thats just one of the big up front costs of going into manufacture. What about the salaries of the engineers who worked for 6+ months to design the chip, will be working x months to ensure it works and continue optimising the design, and presumably will also be working 6+ months on the next generation chip? or are they all expected to work for free? What about the sales staff and the support staff ? the people who answer the phone to help people with orders, or help people with problems getting it running etc? And what about the return on investment for the investors who put up the money to start the company in the first place?
also, in the timeframes youre talking about, to design a 400-600 GH chip that you havnt even started yet, would take you 6+ months to design it, then 3+ months to manufacture it, so you're already 9 months behind the competition. In that amount of time, their chips (which have been dropping in line with the rise in difficulty) will already be the price that youre aiming at, obviating the need for you to do this exercise.
by way of an example look at Cointerra (simply because theyre the lowest cost asic (per GH) i know of at present). they're selling their december systems (not just chips, actual full systems in a box with power supply, controller host, plug and play...) at $7/GH, and their January systems for $3/GH. By the time we get to may, june.. when your chips might be ready, just imagine what price theyd be selling them for!? And what about every other asic company thats also competing for the same business. theyre not in a vacuum. theyre competing not only with each other, which keeps them honest, but theyre also selling gigahashes at a price that people decide has an roi or they wouldnt sell any, so they have to keep dropping their price in line with network hash increases. And all of these guys would prefer to be selling chips on their own instead of systems, and im sure the only reason they build systems is because of time to market issues, and if they left it to others to design their boxes, theyd be out in the marketplace later.
part of the problem that youre missing is that its iots the NRE's that are causing the biggest problem. its the high cost of starting a new asic that creates the issue (and lots of asics compound the problem). not the actual production cost. the bitcoin world actually needs LESS new asics, because its the raisning of the cash for the nre that puts such a strain on the bitcoin economy (all the pre-orders, many of which are paid for in bitcoins, which are converted into fiat and then paid directly to the fab, to pay for the nre). if we didnt have such high nre's, asic companies wouldnt need pre-orders, and the price of bitcoins wouldnt have such price pressure because of high sales of btc into fiat.
youre compounding the problem. Why not instead, do a deal with knc, hashfast or cointerra to buy the rights to their old 28nm chip when theyve obsoleted it (in the timeframe that your one would be ready anyway) and theyve moved on to 20, 16 or 14nm for their next chips ?
-- Jez
the goal of an asic company is not just to sell enough chips to pay for its NRE. thats just one of the big up front costs of going into manufacture. What about the salaries of the engineers who worked for 6+ months to design the chip, will be working x months to ensure it works and continue optimising the design, and presumably will also be working 6+ months on the next generation chip? or are they all expected to work for free? What about the sales staff and the support staff ? the people who answer the phone to help people with orders, or help people with problems getting it running etc? And what about the return on investment for the investors who put up the money to start the company in the first place?
also, in the timeframes youre talking about, to design a 400-600 GH chip that you havnt even started yet, would take you 6+ months to design it, then 3+ months to manufacture it, so you're already 9 months behind the competition. In that amount of time, their chips (which have been dropping in line with the rise in difficulty) will already be the price that youre aiming at, obviating the need for you to do this exercise.
by way of an example look at Cointerra (simply because theyre the lowest cost asic (per GH) i know of at present). they're selling their december systems (not just chips, actual full systems in a box with power supply, controller host, plug and play...) at $7/GH, and their January systems for $3/GH. By the time we get to may, june.. when your chips might be ready, just imagine what price theyd be selling them for!? And what about every other asic company thats also competing for the same business. theyre not in a vacuum. theyre competing not only with each other, which keeps them honest, but theyre also selling gigahashes at a price that people decide has an roi or they wouldnt sell any, so they have to keep dropping their price in line with network hash increases. And all of these guys would prefer to be selling chips on their own instead of systems, and im sure the only reason they build systems is because of time to market issues, and if they left it to others to design their boxes, theyd be out in the marketplace later.
part of the problem that youre missing is that its iots the NRE's that are causing the biggest problem. its the high cost of starting a new asic that creates the issue (and lots of asics compound the problem). not the actual production cost. the bitcoin world actually needs LESS new asics, because its the raisning of the cash for the nre that puts such a strain on the bitcoin economy (all the pre-orders, many of which are paid for in bitcoins, which are converted into fiat and then paid directly to the fab, to pay for the nre). if we didnt have such high nre's, asic companies wouldnt need pre-orders, and the price of bitcoins wouldnt have such price pressure because of high sales of btc into fiat.
youre compounding the problem. Why not instead, do a deal with knc, hashfast or cointerra to buy the rights to their old 28nm chip when theyve obsoleted it (in the timeframe that your one would be ready anyway) and theyve moved on to 20, 16 or 14nm for their next chips ?
-- Jez
