LTC FPGA Miner

[alpha-t]

Hello Everyone, this is my first forum post. As an electrical and electronic engineer based in the U.K, i have been very interested in LTC FPGA mining, and in my detailed analysis of the Scrypt c-code it is very doable. I have actually started design and have extensive IP capability with both memory controllers and cryptography. The Scrypt based proof-of-work is now familiar to me after analysis and i am working with a large reputable Asian company, using my personal funds to building a cost effective FPGA for consumer use. I will be posting both prototype images and videos as development continues (showing proof of hashing). A release of a website and more information regarding hashing speeds, low power consumption, scalability and ease-of-use. I cannot disclose the company i'm working with at the moment, but i will at the prototype stages and also a link to the press-release from the company itself. But they are one of the world's most experienced in production of cryptography and security processing hardware. At a later stage YOU the community itself can take part in the whole process, as further knowledge specifically in the calibration of the mining software which would run on the host computer,  MAY be needed. Thank You.

[trebor_trebor]

It is not possible to mine LTC on existing FPGAs ?

[trebor_trebor]

Good Luck and post something soon

[gingernuts]

Sounds interesting - what sort of hash rates are you aiming for?

[jpyao78]

interesting

[diamondp]

what kind of hashrate, cost per unit, and power consumption do you expect?

[Lokes]

Sounds interesting. I can't wait to see the details.

[ilostcoins]

I'm hoping FPGA won't be much more efficient than GPU. More specialized equipment will narrow down participation, which is an important factor for wider adoption.

[Smiley_01]

I fear this will be a bit more expensive then GPU (per Hashrate).
Mainly because of the expensive high speed mem interface (max. DDR3 with FPGA currently afaik).
And the FPGA with a lot of on chip mem are very expensive. Have done some small research too, doen't
look to well currently...

There have been some discussions and calculations about this already - either in the altcoins subforums
or at litecointalk, can't seem to find it currently...

[bebopbob]

I wonder when a graphics cards company will release a gpu that is specifically optimized for mining.  Something like the 6880 BTC edition.

[rammy2k2]

u cant mine with FPGA since FPGA has no RAM ...

[NaHoW]

You can modify FPGA cards and add ram. But then you would probably be cheaper of just selling modified GPU's.

[WindMaster]

u cant mine with FPGA since FPGA has no RAM ...

Depends what you're referring to when you say FPGA:
http://en.wikipedia.org/wiki/Fpga

I've noticed a large number of people in the cryptocoin community misuse the term, apparently using it to refer to an end-user product designed for mining that happens to contain an FPGA.  But the circuit board as a whole is *not* an FPGA, the FPGA is a chip that's located on that board.

All FPGA's have RAM.  You can't have a CPLD block at all without at least some D flip-flops (which would store 1 bit of memory), and an FPGA is made up of a *huge* number of CPLD blocks.  And then, most modern FPGA's also have some quantity of dedicated on-die SRAM too.  Including the Xilinx Spartan-6 FPGA's that were used in many of the FPGA-based mining products released to date.

[WindMaster]

You can modify FPGA cards and add ram. But then you would probably be cheaper of just selling modified GPU's.

By far.  The reason GPU's are useful for mining scrypt-based coins is that their memory bandwidth is very high.  You just can't compete with the memory bandwidth of modern Radeon GPU's with any of the commercially available FPGA's when it comes to the memory throughput requirements of scrypt.

[WindMaster]

It is possible. However, you will see very small hashing rates as the current FPGA's are designed for Bitcoin Mining and are very CPU intensive rather than memory intensive. So custom FPGA memory interfacing would be required. Keep a look out, i will be giving more specific information in the coming days. I know actions speak louder then words, so i will not post anything more till we have video/image concrete proof backing up my claims. Thank You

I think a better approach may be to not mention you're working on FPGA scrypt implementation until you actually have something to show.  Having gone down this path and actually having implemented scrypt in FPGA's (Xilinx Artix-7 specifically) with a variety of theoretical and attempted optimizations, you pretty much have to come up with some revolutionary way to shortcut the way scrypt is calculated before you'll have any chance of an FPGA implementation remotely approaching the price/performance ratio of GPU's.  That goes even for the possible power advantages, unless someone is going to mine for a decade with the same hardware to achieve ROI on power costs.  And if you find a revolutionary shortcut to calculating scrypt, it's likely that same technique will apply to calculating scrypt on GPU's instead, and far more money will be derived by doing so and mining LTC for all it's worth with a more optimal GPU implementation than by scamming investors with pre-orders for FPGA-based mining products.

Observe the other FPGA threads on here, where people initially state they have no intention of taking pre-orders or investors and aren't going to reveal any details about just how they're going to shortcut scrypt calculations.  Then look a few pages later where people are literally waving gobs of cash to that person pleading to be first in line (even when everyone is agreeing the cost/performance ratio will be worse than GPU's), only to have the person reluctantly agree to accept investors.  It's all played out with glorious theatrics to make it look like it was the investors that had the idea to throw money at the developer rather than the other way around..

[alexey8444]

Oh, it's very interesting and actual now for scrypt-cryptocurrencies 

[J35st3r]

For all your words, I don't actually see any technical detail here eg FPGA family, external ram bandwidth.

Your competitors at http://www.litecoinfpga.com/ have been much more forthcoming. I wish you well, but at the moment its all just vaporware from a totally unverified source. The scamhammer is poised and waiting 

[J35st3r]

Thanks for your reply. After viewing the site you gave me and the discussed thread regarding our 'competitor' there is no indication of any specifications AT ALL. And it seems they are looking for manufacturers to contact them regarding pcb fab. So i do not classify them as a competitor at all.

And as iv'e stated before we are just getting our intentions known and are far away from any pre-orders. Besides from technical details, a press release from our partner will be shown and a prototype demonstration. The next stage will be to actually show the implementation on an off the shelf board, which hasn't even been shown as proof of concept as of yet by anyone. Along with which FPGA + board configuration we will be going ahead with, which i stated in my previous post.

As i'm quite busy ill keep my responses down to comments that i feel require replying to and any updates on our progress.

Once again thanks for your interest and maybe you should not be so quick to use your 'scam hammer'.

Thanks for the reply. I'm a little surprised that you have not seen Jasinlee's work as its pretty much the top item for Google search on Litecoin FPGA (and he's been busy on this for several months now). It does seem that you have not been very thorough in your research to date.

The scam hammer is merely poised, but this being the newbie forum, I do like to give the benefit of the doubt on the scam-side (I was a newbie once myself, and caution is to be welcomed)