I am carefully considering all sides, but my engineers, while capable engineers, are not economists, so if my customers can recoup part of their investment into equipment after it's useful life for bitcoin by using the equipment for something else or by selling just the processor or if I can lower per chip cost by making Multi-purpose / Multi-project chip and selling 5 times the number of chips at the expense of some performance of each single chip - it should at least be explored. I know that some may think that it insane, but they can't see perspectives that I can.
If it's additional project then I can disable enough SHA cores after it is made and before it is sold, so it still can do SHA fast but not fast enough for bitcoin , so no one say they buy them for something else and then go and use them for bitcoin.
But if it's additional hashing protocols is there a market for hundreds of thousands chips with crazy fast performance ? A lot of chip manufacturers integrate slow sha into their own chips. What other uses are there for a massive hashing chip?
I mean small once could be put everywhere like a network security device, but mine is a big chip. I still have some time to change it into smaller one. Making them very small will increase ability to be sold for use in other products but will increase cost a bit for making high density bitcoin system, but if they later can be sold for more than the small increase in cost I think it should be considered.
Yes, this semiconductor market is new to me, but I am learning fast and trying to innovate.
Topic: Multi-purpose / Multi-project chip ? (Read 786 times)
Okay, despite the accusations of scam in your other thread, I'll throw out some suggestions even if it is only for the benefit of others. As some would say, I'll bite.
What I'd like to see in an ASIC is not just a Bitcoin only application. I'd like to see a generic SHA ASIC. So, say it supports SHA-256 natively for Bitcoin, but you could maybe also do SHA-384 or SHA-512, etc. Maybe even some other hashing algorithms if they share enough in common.
However, you can't really make it an all-in-one chip either, otherwise you're practically making a CPU and losing the ASIC advantage. So you'd have to balance supporting other hashing algorithms without bloating transistor count, etc. But if you could make an ASIC that does more than just SHA-256, you could have a lot of security uses. Admittedly some might say they'd only be used for blackhat scenarios (i.e. creating rainbow tables), but that's another discussion in itself.
I don't see any scrypt support being a good idea. That requires too much expensive cache and the utility and value of altchains that use scrypt would be debatable for many Bitcoin miners.
But basically too much of a deviation from what makes just SHA-256 tick will just increase the ASIC's cost while lowering its potential hashing capability. Of course, you should mostly be consulting what your engineers are saying... I don't design ASICs and probably only a few here do.
What I'd like to see in an ASIC is not just a Bitcoin only application. I'd like to see a generic SHA ASIC. So, say it supports SHA-256 natively for Bitcoin, but you could maybe also do SHA-384 or SHA-512, etc. Maybe even some other hashing algorithms if they share enough in common.
However, you can't really make it an all-in-one chip either, otherwise you're practically making a CPU and losing the ASIC advantage. So you'd have to balance supporting other hashing algorithms without bloating transistor count, etc. But if you could make an ASIC that does more than just SHA-256, you could have a lot of security uses. Admittedly some might say they'd only be used for blackhat scenarios (i.e. creating rainbow tables), but that's another discussion in itself.
I don't see any scrypt support being a good idea. That requires too much expensive cache and the utility and value of altchains that use scrypt would be debatable for many Bitcoin miners.
But basically too much of a deviation from what makes just SHA-256 tick will just increase the ASIC's cost while lowering its potential hashing capability. Of course, you should mostly be consulting what your engineers are saying... I don't design ASICs and probably only a few here do.
Chip manufacturing is expensive and chips become obsolete very fast so I would like to amortize NRE over larger production runs of chips and maybe give them second life after their primary function is surpassed by even newer chips.
So to that end while I have advice from my engineers I would like to crowed source what other functions or projects can I put into chip to give it a longer life and better resale value after bitcoin have moved in difficulty.
So to that end while I have advice from my engineers I would like to crowed source what other functions or projects can I put into chip to give it a longer life and better resale value after bitcoin have moved in difficulty.
Jump to: