Author Topic: Fibonacci.io Scrypt ASIC - Questions & Info. - Its bigger on the inside.  (Read 1130103 times)

Offline digitalindustry

  • Jr. Member
  • **
  • Posts: 24
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #45 on: April 29, 2013, 01:03:26 PM »
How much are you looking for capital wise for an initial investment?

Offline rriky92

  • Newbie
  • *
  • Posts: 16
    • Riccardo Rossi's Website
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #46 on: April 29, 2013, 02:49:24 PM »
This is a great news!

Offline jasinlee

  • Scammer
  • Elite Member
  • ************
  • Posts: 1549
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #47 on: April 29, 2013, 04:23:02 PM »
Capital, I am not sure yet. I am actually working on that right now.

Offline yxxyun

  • Newbie
  • *
  • Posts: 4
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #48 on: April 30, 2013, 08:37:38 AM »
are you chinese ? and where do you plan to sell your stock? [LTC-GLOBAL] Litecoin Global Exchange website?

Offline dan99

  • Newbie
  • *
  • Posts: 9
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #49 on: April 30, 2013, 09:48:51 AM »
maybe he is chinese, by the name of lee at the end of his name... :) but in china it can be gotten really.. yes really .. i travel there very often and got good connections my ancestors hometown  :) we speak the same tongue so to saY.

Offline Lady_ZENITH

  • Newbie
  • *
  • Posts: 14
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #50 on: April 30, 2013, 12:33:51 PM »
Hmm Iv got a question....

And that is " how do you actually plan to do it? "

It is principle of scrypt to be bottlenecked by memory bandwith, thus incompatible with existing FPGA or asic chips.  So, hear my logic...

The current fastest device is AMD Tahiti chip which is with its bus bottlenecked by memory bandwith more then the hashing performance of the GPU itself. 7Ghz 384-bit GDDR5 are bottlenecking the operation at about 700Kh/s. That kind of bandwith limitation we have here now.

Existing FPGA's when it comes to memory bandwidth are dreadful. They are just horrible. I cannot imagine any of them being of any use. They were usable for Sha256 for obvious reasons, but that same thing cannot be programmed to mine scrypt effectively cause it does not have the needed bandwidth.

The only way how I imagine that being done is some kind of large and fast on-die cache. Which would result in large chip that has to be made using very advanced manufacturing process, like 28nm.  Or.... again big chip that would compensate it with wide memory bus and fast modules, again, large and expensive to manufacture.

The companies that make the asics for SHA256 seem X-times bigger, yet they cannot afford such technology. Avalon is made on the ancient130nm (2001/2 technology) and the promised and forever delayed BFL stuff is expected to be old 65nm (2006 stuff). Both of what they use is outdated and not much usable for the chip we would need.  So... just... how. How do you plan to design such complex thing, get money for that, and then also compete and pay for the advances silicon manufacturing, when the old big guys in FPGA and asic cannot offer that themselves. I have no idea.... 



Offline jasinlee

  • Scammer
  • Elite Member
  • ************
  • Posts: 1549
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #51 on: April 30, 2013, 01:13:45 PM »
Magic..... jk well I cannot disclose it yet but we are not using the normal channels to achieve our goals....or the normal design most would use.

I am the whitest white boy you will ever meet just fyi.

Offline dan99

  • Newbie
  • *
  • Posts: 9
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #52 on: April 30, 2013, 02:19:48 PM »
Hmm Iv got a question....

And that is " how do you actually plan to do it? "

It is principle of scrypt to be bottlenecked by memory bandwith, thus incompatible with existing FPGA or asic chips.  So, hear my logic...

The current fastest device is AMD Tahiti chip which is with its bus bottlenecked by memory bandwith more then the hashing performance of the GPU itself. 7Ghz 384-bit GDDR5 are bottlenecking the operation at about 700Kh/s. That kind of bandwith limitation we have here now.

Existing FPGA's when it comes to memory bandwidth are dreadful. They are just horrible. I cannot imagine any of them being of any use. They were usable for Sha256 for obvious reasons, but that same thing cannot be programmed to mine scrypt effectively cause it does not have the needed bandwidth.

The only way how I imagine that being done is some kind of large and fast on-die cache. Which would result in large chip that has to be made using very advanced manufacturing process, like 28nm.  Or.... again big chip that would compensate it with wide memory bus and fast modules, again, large and expensive to manufacture.

The companies that make the asics for SHA256 seem X-times bigger, yet they cannot afford such technology. Avalon is made on the ancient130nm (2001/2 technology) and the promised and forever delayed BFL stuff is expected to be old 65nm (2006 stuff). Both of what they use is outdated and not much usable for the chip we would need.  So... just... how. How do you plan to design such complex thing, get money for that, and then also compete and pay for the advances silicon manufacturing, when the old big guys in FPGA and asic cannot offer that themselves. I have no idea....

Hi Lady Zenith,

Could you make some suggestions on how to create a good Litecoin or Bitcoin Miner, like those using 28mnm technology? Jasinlee is not perfect
however he is working on something though.. I don't know where u are located .. Avalon even though is consider Old Tech and make in the far east..  are a functioning and kicking compares to butterfly. :)

Offline Lady_ZENITH

  • Newbie
  • *
  • Posts: 14
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #53 on: April 30, 2013, 06:20:46 PM »
Yep cause it is asic, few transitors, one single purpose.

That purpose which scrypt is denying by its principle. To mine scrypt you need bandwith. I simply do not know about any PGA with strong bandwith. So this is why I do not trust this thing without any sort of design ideas or such stuff introduced.  Since that is being held secret and I cannot imagine a way how to afford making it without being a silicon manufacturer itself (UMC, TSMC, etc.).... you draw your own conclusion.

Offline Magnate

  • Jr. Member
  • **
  • Posts: 65
    • Bling My Bike
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #54 on: April 30, 2013, 07:27:33 PM »
If you did make an FPGA on even 45Nm with plenty of DDR5 and high band width, you would just sell it to the litecoin market, I'm sure there would be a wider market interested in it.

However since the current FPGA manufacturers don't make such a chip suggest the market isn't that huge- they make what their current customers what, which is just a flexible chip that is cheaper than asic but faster than a less specific chip. On board RAM doesn't seem to be a market requirement.

But if you did a business plan for a litecoin ASIC, I would actually step back to FPGA simply because of the wider market you could sell to with a litecoin FPGA. I can see the potential for huge success with a good FPGA with high band width ram.

Offline jasinlee

  • Scammer
  • Elite Member
  • ************
  • Posts: 1549
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #55 on: April 30, 2013, 10:25:30 PM »
DDR5 is not even necessary, we figured that out pretty early in the project. Its also more expensive to purchase on a large scale than DDR3.

Offline Malak

  • Sr. Member
  • ****
  • Posts: 570
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #56 on: April 30, 2013, 10:58:35 PM »
Hope this works out. I'm mainly replying so I can get notifications of updates.  lol

Offline thesnoo23

  • Newbie
  • *
  • Posts: 10
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #57 on: April 30, 2013, 11:06:29 PM »
pff. notifications of updates? I just camp the thread :P

Offline jasinlee

  • Scammer
  • Elite Member
  • ************
  • Posts: 1549
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #58 on: April 30, 2013, 11:13:36 PM »
Oh I should have good juicy info for everyone on Thursday.

Offline thesnoo23

  • Newbie
  • *
  • Posts: 10
Re: LTC FPGA Development - In Progress - News and Questions
« Reply #59 on: May 01, 2013, 12:56:23 AM »
Oh I should have good juicy info for everyone on Thursday.

AAAAAAAHHHHHHH!

You are evil.