*SCRYPT FPGA* - CryptoIndustries.com - official Bitcointalk thread

[c1010010]

i doubt these fpgas will beat undervolted 7970 by much at all, maybe some weak amount like 30% and yet cost 3x the price. Not worth it.

Agreed.

The whole point of SCRYPT was to hardware proof against ASICS.

[Chronikka]

i doubt these fpgas will beat undervolted 7970 by much at all, maybe some weak amount like 30% and yet cost 3x the price. Not worth it.

Agreed.

The whole point of SCRYPT was to hardware proof against ASICS.

Which is ridiculous because asic's can mine scrypt too. The reason people say scrypt is "asic proof" is because the scrypt algorithm requires a good deal of high speed memory which most asic's don't have. A dedicated individual or company could produce an asic based scrypt miner if they design it to provide sufficient memory. We are years away from that most likely. The asic market is just kicking off with bitcoin so don't expect to see scrypt fpga's or asic's for some time

[dragon695]

i doubt these fpgas will beat undervolted 7970 by much at all, maybe some weak amount like 30% and yet cost 3x the price. Not worth it.

Agreed.

The whole point of SCRYPT was to hardware proof against ASICS.

Which is ridiculous because asic's can mine scrypt too. The reason people say scrypt is "asic proof" is because the scrypt algorithm requires a good deal of high speed memory which most asic's don't have. A dedicated individual or company could produce an asic based scrypt miner if they design it to provide sufficient memory. We are years away from that most likely. The asic market is just kicking off with bitcoin so don't expect to see scrypt fpga's or asic's for some time

Providing sufficient memory that can be accessed in a timely manner is what makes GPUs cost the price they do. Also, as you increased memory, your savings on power consumption diminish. The problem isn't that you can't mine scrypt with ASICs, the problem is that the efficiency gains of doing so are negligable at best, subpar at worst. People lucked out with SHA256, ASICs could be designed that gave significant efficiency gain. If you check out the FPGA situation, it really isn't looking promising.

[Chronikka]

i doubt these fpgas will beat undervolted 7970 by much at all, maybe some weak amount like 30% and yet cost 3x the price. Not worth it.

Agreed.

The whole point of SCRYPT was to hardware proof against ASICS.

Which is ridiculous because asic's can mine scrypt too. The reason people say scrypt is "asic proof" is because the scrypt algorithm requires a good deal of high speed memory which most asic's don't have. A dedicated individual or company could produce an asic based scrypt miner if they design it to provide sufficient memory. We are years away from that most likely. The asic market is just kicking off with bitcoin so don't expect to see scrypt fpga's or asic's for some time

Providing sufficient memory that can be accessed in a timely manner is what makes GPUs cost the price they do. Also, as you increased memory, your savings on power consumption diminish. The problem isn't that you can't mine scrypt with ASICs, the problem is that the efficiency gains of doing so are negligable at best, subpar at worst. People lucked out with SHA256, ASICs could be designed that gave significant efficiency gain. If you check out the FPGA situation, it really isn't looking promising.

I agree which is why I said we are years away. Somebody will build one when parts are less expensive and easier to come by. Let sha256 establish the market for newer, mass produced asic technologies like it has been doing, and in a few years a scrypt fpga/asic is more probable. In the meantime this is an interesting project

[jdebunt]

some information regarding the recent q&a with The Koolio

http://majesti.co/cryptonerd/crypto-industries-qa-about-their-scrypt-fpga/

[fattypig]

i doubt these fpgas will beat undervolted 7970 by much at all, maybe some weak amount like 30% and yet cost 3x the price. Not worth it.

Well, once they are done with the current design, they will surely increase it furthermore like making it more energy efficient, cheaper and hash faster.

[BitcoinEXpress]

some information regarding the recent q&a with The Koolio

http://majesti.co/cryptonerd/crypto-industries-qa-about-their-scrypt-fpga/

Koolio said....


~BCX~

[nightengale]

Will you accept escrowed funds for purchases?

[PrintMule]

But... but...

Blasphemy!

[markm]

some information regarding the recent q&a with The Koolio

http://majesti.co/cryptonerd/crypto-industries-qa-about-their-scrypt-fpga/

So they are putting some kind of "it is our machine not your machine" crap on it to prevent it mining coins they do not "authorize" or something like that

Taking a percent too, maybe? Maybe forcing you to have to use their pools thus not be able to mine coins they choose not to support

TL;DR How the heck do they plan to prevent people using them to mine whatever coins the user chooses to mine, and WHY

-MarkM-

[6strings]

Seriously? Is any of this actually the case?
I was honestly planning on picking one of these up(assuming price was right) but certainly wont if any of the above is true.

[nightengale]

Seems like you're trolling a bit, I didn't read that stuff in the article nor did I anywhere else.

I actually received a personal response on my inquiry about escrow -- a nice touch. I'll be watching this one.

[int3ractivodular]

Seems like you're trolling a bit, I didn't read that stuff in the article nor did I anywhere else.

I actually received a personal response on my inquiry about escrow -- a nice touch. I'll be watching this one.

If escrow is accepted that's definitely a good thing, but if they actually plan on limiting customers to only certain coins, LOL NOPE. 

[narousberg]

awaiting prices

[Totscha]

Since nobody wants to tell anything about prices, I'll give you my very (repeat: very) rough estimate.

Frist thing. We know they will be using Atrix-7 200T in a 1156 pin package FPGA. Motaboy (the designer) says they have 104 kH/s per chip (theoretical). Reference:  (https://forum.litecoin.net/index.php/topic,6211.30.html)

ONE chip costs about 250-300$. For the 1MH/s unit they will need at least 10 chips, 12 more likely. We are at $3500 now.

We don't know much about the other hardware (case, boards, RAM, PSU), but it could be done for about $500. So about $4000 / MH/s.


A 2 MH/s GPU rig will set you back about $2000.


Basicaly their FPGA will give 4x LESS hash power per $ investment. What is interesting is the lower power consumption.

Profitability at current Litecoin diff prices, etc:
GPU (2MH/s, $2000, 1000W, 0,15/kWh): break even in 218.90 Day(s)
FPGA (1MH/s, $4000, 250W, 0,15/kWh): break even in 738.28 Day(s)

So what they need to do is get the chips at at least 50% discount to get even close to GPU mining...

[karlb]

Can we point the device at a pool of our choice or are we restricted to one pool, or a handful of pools?

[Simran]

http://thecryptoblog.com/ask-cryptoindustries-questions/

Basically, Koolio from CryptoIndustries and I have decided on a small event(Should I say that?) where you ask a question in the comments, both Koolio and Motaboy will read them and send me their responses and I will go ahead and make another blog post with the questions and answers!
This event will be from November 25th to December 2nd, and will be answered between the 2nd and the 3rd! Any updates will be made here, so stay posted!

You can ask questions like, “When are the Litecoin FPGAs going to come out?” or possibly “Any plans for a Litecoin ASIC?” and so on. Please keep the questions mature, or they won’t be answered or I will delete them and block you from Disqus.

Be aware that not all questions will be answered, but they will try to answer each one appropriately. If your answer isn’t answered, I will try to host future events where you can ask again!

To ask a question, simply comment below. If you are unable to ask a question using Disqus, please either send me an email using the Contact form above, or comment on the Reddit/BitcoinTalk listing. If all fails, just message me on BTC-e or find me on the IRC.

Let the questioning begin!

[int3ractivodular]

 “When are the Litecoin FPGAs going to come out?” or possibly “Any plans for a Litecoin ASIC?”

[sl877]

Whats the prices for the 3 miners!

[Paladin69]

Since nobody wants to tell anything about prices, I'll give you my very (repeat: very) rough estimate.

Frist thing. We know they will be using Atrix-7 200T in a 1156 pin package FPGA. Motaboy (the designer) says they have 104 kH/s per chip (theoretical). Reference:  (https://forum.litecoin.net/index.php/topic,6211.30.html)

ONE chip costs about 250-300$. For the 1MH/s unit they will need at least 10 chips, 12 more likely. We are at $3500 now.

We don't know much about the other hardware (case, boards, RAM, PSU), but it could be done for about $500. So about $4000 / MH/s.


A 2 MH/s GPU rig will set you back about $2000.


Basicaly their FPGA will give 4x LESS hash power per $ investment. What is interesting is the lower power consumption.

Profitability at current Litecoin diff prices, etc:
GPU (2MH/s, $2000, 1000W, 0,15/kWh): break even in 218.90 Day(s)
FPGA (1MH/s, $4000, 250W, 0,15/kWh): break even in 738.28 Day(s)

So what they need to do is get the chips at at least 50% discount to get even close to GPU mining...

Thanks for posting.  This must be why there is no competition.  Price doesn't make sense yet.  It'll probably get there.