Topic: [ANN] Fibonacci Info and questions - page 19. (Read 104190 times)

I woudn't even consider letting these things out into the wild until viable competing devices show up.

If you folks truly have what you say you have, i hope you have a serious fab contract. 

Ok lets chill, it will only be in ltc.

you could be so smart 2 post ago then there would be any conversation about it 

Nope. Not in a long time. Several years ago. (My forum nick has nothing to do with pot, it's a coincidence.)

I know he said LTC, but maybe they might give a discount to those who will pay with BTC. If not, then I will simply buy LTC using my BTC and pay that way. The miners will still be pegged to it's fiat cost.

lol, if he said LTC that means LTC not BTC or other coin. you smoke pot?

What would you need to ship immediately instead of host? I'm assuming if you can trust who you send to, to not reverse engineer it. Like, for example, uh, escrow agents? hehehe.

*edit* Also, I won't sell it to anyone else until you start shipping it to the rest of the world.

And, does your discount only apply to LTC? Can we get the same discount for BTC? (For those who don't want to spend their LTC, or for those who don't have LTC, to prevent them from having to buy on an exchange.)

Thank you for the clarification windmaster.

That is good then no one will be upset that we dont display it to the world 

We know what we have, there are a few community members that are aware too. All of them agreed this was the best approach, it was actually suggested by one of the forum admins in the first place to host and not ship immediately.

Just playing devil's advocate here, but I'd tend to say that a team capable of reverse-engineering an ASIC by selectively etching layers and taping-out a replica, already has the skillset to just develop an scrypt core from scratch.  Given that scrypt's internal workings aren't secret knowledge, and the possible hardware approaches to go about it are generally self-evident and well-documented from scrypt research outside of the cryptocurrency community, I think it would take less time to implement from scratch rather than reverse someone else's design.  If we were talking about replicating, say, Sandy Bridge, or some of AMD and Nvidia's GPU dies, that would be one thing.  But an ASIC implementation of scrypt just isn't that hard to do from scratch.  95% of other cores that end up in various ASIC's across other industries are significantly more complex than an scrypt core.

I'd say any team with the knowledge needed to reverse an ASIC, can probably also easily determine the appropriate TMTO and SRAM size (or external DDR3/GDDR5 size) for the particular silicon process node they're targetting to optimize cost/performance (and, really, there's only 2 choices along the TMTO spectrum that are going to land near the peak cost/performance for an ASIC implementation with on-die SRAM, and only 1 optimal TMTO choice for external DDR3 or GDDR5).  I don't see where there's much to gain in anyone's ASIC development cycle from looking at someone else's scrypt die.

Negative.  The logic would be radically different between a Litecoin-oriented Salsa20/8 (1024,1,1) implementation and a YACoin-oriented ChaCha20/8 (N,1,1) + Keccak512 implementation.  There is very little similarity between the scrypt variant used in Litecoin and the scrypt variant used in YACoin.  Additionally, YACoin uses variable N while Litecoin has fixed N=1024.  For YACoin, N is currently 32768, and will change to 65536 on 31 May 2014.  Details on the variable N schedule can be found in my post here:

https://bitcointalksearch.org/topic/m.2162620

When YACoin launched at N=32, I adapted 4 of our earlier FPGA LTC prototypes (populated with XC7A200T's and a pile of DDR3) to YACoin instead.  It was real easy pickings at N=32.  At N=64, not so much.  At N=128, I stopped and moved back to a different more efficient hardware solution for YACoin mining.

Variable N would make ASIC implementation a real bitch, since you can't optimize the mixing function's logic for a specific value of N like you can with Salsa20/8 (1024,1,1).

It is, and I know it would take someone months or even 1+ years to reverse engineer it. But I know what our teams values are, and I know (after dealing with lots of potential VCs) other companies have different values. So I feel more comfortable making them work hard to beat/meet our numbers rather than giving them the answers.

I guess that means the same ways we all use for reverse engineering of ASICs.

That is a cruel world we live in, isn't it?

I may have been awake over 24 hours when I wrote that....I apologize.

I just realized that I put in a feel felt found in there lol.

LMAO! - That is awesome...

And picture is like TELLING thousand words. I've seen "pictures" of SHA-scrypt combined ASICs and nothing after...

Been a bit busy, I will ask my engineer to be sure. But looking at what jane is (Chacha) it probably would yes.

Sorry if this has been asked before, but will it work with Scrypt-Jane, such as YaCoin?
Thanks,
Flippyfeet

There are several ways to tell the difference in the type of miner you are using.

FPGA scrypt mining was known already. It was less profitable in comparison to GPUs. With Your intention of not selling "scrypt ASIC" one can start being suspicious that at the beginning You will sell expensive shares of actualy GPU farms.

i am in both to buy the hardware and the IPO, keep me posted