Author Topic: Alpha Technology Litecoin (Scrypt) ASIC Miner Order Batch 1 Now!  (Read 360911 times)

Offline icfpgachn

  • Newbie
  • *
  • Posts: 9
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #45 on: October 14, 2013, 04:22:39 PM »
the power efficient is an important parameter.
any normal CPU miner  power efficient will not get 5khs/W,no matter arm,x86,powerpc,MIPS,etc.
any fpga miner will not reach the power efficient limit 15khs/W with internal ram,nor 10khs/W with external ram,no matter DDR,DDR4,GDDR5.
LTC ASIC   power efficient will not get 50khs/W.
so,if you see a miner with  power  efficient over the limit,it is a scam.
« Last Edit: October 15, 2013, 12:28:48 AM by icfpgachn »

Offline brib

  • Jr. Member
  • **
  • Posts: 22
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #46 on: October 14, 2013, 05:17:16 PM »
Jk_14.. I think your referring to FPGA this is not an FPGA project anymore, it is ASIC.. which a lot of people said would happen instead of fpga because of fpga having limitations, jasinlee seems to disagree though with their fpga with no proof of people he is workin with or anything. And is accepting money alrdy!

AFAIK there is no way to build an ASIC with no FPGA prototype.

Offline kramble

  • Full Member
  • ***
  • Posts: 247
    • Github
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #47 on: October 14, 2013, 05:34:11 PM »
Jk_14.. I think your referring to FPGA this is not an FPGA project anymore, it is ASIC.. which a lot of people said would happen instead of fpga because of fpga having limitations, jasinlee seems to disagree though with their fpga with no proof of people he is workin with or anything. And is accepting money alrdy!

AFAIK there is no way to build an ASIC with no FPGA prototype.

Simulation. Its been done that way for decades. (And in the old days before FPGAs the prototypes were built out of MSI parts (74/4000 series) and wire-wrap).

Not that I'm condoning Alpha Techonology's project. ASIC just does not make sense given the current market-cap of LTC, the NRE is way too high. This project is going nowhere.

Offline dunebuggy

  • Newbie
  • *
  • Posts: 19
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #48 on: October 14, 2013, 05:53:09 PM »
If i have the option to buy something cheaper and faster (ASIC) why would i pay for an FPGA at $1.50/khash?  It makes perfect sense for me. From what i understand i dont know about ddd3 etc, maybe that wouldnt be required in an ASIC , e.g. if you can get e.g 50 khash from 8 cores, an ASIC having hundreds of cores per ASIC chip, would ddr3 interfacin rlly be required? I would like Krambles opinion on this too.

I suspect there is much work to be done from them, i doubt them figures are final. Making an Asic takes so long. Will just wait to see how this develops and who they r working with and their fpga prototype etc
« Last Edit: October 14, 2013, 06:02:27 PM by dunebuggy »

Offline Ballistix

  • Full Member
  • ***
  • Posts: 219
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #49 on: October 14, 2013, 06:01:12 PM »
Damn, 5000 khash for 200w..... The thing I liked about scrypt coins was the apparent asic proof design of them. Now if this all legit and they start shipping as promised when they promised, we will see skyrocketing dificulty, once again pushing gpu's out. Full on expedited race to the bottom and the ones actually getting rich are those at alpha technology.

Offline Xingqiaoyin

  • Full Member
  • ***
  • Posts: 433
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #50 on: October 14, 2013, 06:12:11 PM »
Ah yes the ASIC race, where the first adopters are rewarded with an ROI of weeks and will never see ROI if they started 2 months too late.
So I just put in the numbers 5000 khash/200watt @12c in burnside calculator.
Profit per month is 275 usd.
Seeing how fast ASIC obseletes, the most I'm going to pay for a preorder is 550 usd.
Let's see how attractive that price is for ALphaTechnology.

Offline kramble

  • Full Member
  • ***
  • Posts: 247
    • Github
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #51 on: October 14, 2013, 06:17:10 PM »
If i have the option to buy something cheaper and faster (ASIC) why would i pay for an FPGA at $1.50/khash?  It makes perfect sense for me. From what i understand i dont know about ddd3 etc, maybe that wouldnt be required in an ASIC , e.g. if you can get e.g 50 khash from 8 cores, an ASIC having hundreds of cores per ASIC chip, would ddr3 interfacin rlly be required? I would like Krambles opinion on this too.

I don't have any insider-knowledge of the ASIC market, so this is just lay comment, but I'm going to speak my part anyway. The problem is NRE (design and mask costs) which can range from perhaps $50k for a low-spec 130nm process to multiple millions for the sort of high end processes that current GPU s are built on. And an LTC ASIC has to compete with the latter, not the former. So if we ignore the MPW shuttle (which is only suitable for prototyping and university research projects, not volume production), that just leaves hardcopy FPGAs as being a possible contender here. These won't give the sort of performance level that could be expected from a true standard-cell or custom ASIC, but will improve on the current FPGA designs (jasinlee, koolio et al). I can only assume that this is what Alpha-Technology have in mind, so perhaps there is a product in the pipeline (though bear in mind the lead time here, jasinlee and koolio have spent months designing their products, and here comes Alpha-Technology with a more difficult solution, and they couldn't even get their FPGA design to perform at anywhere near its potential).

On the actual design itself. A true ASIC could contain sufficient on-chip RAM to be fully self-contained. If its using conventional off-chip DRAM then it has exactly the same problem as the FPGA designs and is limited by how many DRAM controllers and I/O ports can be fitted. I cannot honestly see how either solution can compete on cost with the current GPU video/mining cards, let alone the next generation. As for power consumption, ipfpgachn has posted above, and I'll defer to his expertise.

... and I've jabbered on far too long as usual and there are two intervening comments. POST.

Offline AlphaTechnology

  • Jr. Member
  • **
  • Posts: 42
    • Alpha Technology
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #52 on: October 14, 2013, 06:53:16 PM »
Been reading some of these comments. In response to Kramble, you are probably referring to the proof of concept (basic implementation) we released in a report. The goal of that was just to show the feasibility if we will be going ahead with an FPGA as a final product, which we are not. So it did not contain any optimisation.

In order to come up with a pre-asic implementation we implemented a single co-processing unit in FPGA at best of its throughput meaning things like pipelining, optimum memory utilization etc. In ASIC this will be multiplied in the 100s. We had minimum targets which we reached.

We will be releasing a video of our FPGA (ASIC) prototype soon and some details of core hashing speed etc. Just for those people requiring more details.

There is more work to be done, and speed/wattage is not final, but it will be around them figures we quoted.

We hope to put everyone's suspicions to an end in the build up to our official launch and pre-order date.

Offline kramble

  • Full Member
  • ***
  • Posts: 247
    • Github
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #53 on: October 14, 2013, 07:06:01 PM »
(duplicate post, see below)
« Last Edit: October 14, 2013, 07:21:17 PM by kramble »

Offline kramble

  • Full Member
  • ***
  • Posts: 247
    • Github
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #54 on: October 14, 2013, 07:17:17 PM »
In order to come up with a pre-asic implementation we implemented a single co-processing unit in FPGA at best of its throughput meaning things like pipelining, optimum memory utilization etc. In ASIC this will be multiplied in the 100s. We had minimum targets which we reached.

That's going to be one very large ASIC if its using hundreds or cores (and 500kH/s is rather on the low side performance-wise for that kind of specification). What process node are you targeting?

I don't mean to be rude, but I don't think you are in OrSoC (KNC)'s league. Last I saw you were a two man team, yourself in Manchester and your developer, Amit Sinha in India. Could you elaborate on how you've expanded your team to support this project, and which partners you have contracted with?

Company details ...
Alpha Technology (Int) Ltd
64 Dickenson Road
Manchester
M14 5HF
Companies House Registration Number:   8483921

That's a residential address here
« Last Edit: October 14, 2013, 07:21:39 PM by kramble »

Offline AlphaTechnology

  • Jr. Member
  • **
  • Posts: 42
    • Alpha Technology
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #55 on: October 14, 2013, 07:25:53 PM »
You just mentioned the head of engineers for our partner company, which was signed as his name on the report we released. If your good at using google, maybe you can figure out who they are, and what kind of expertise they have. That's not a residential address that's our accountancy firm. You do know registered address and trading address are two different things. Anything else?

More information coming soon!
« Last Edit: October 14, 2013, 07:35:40 PM by AlphaTechnology »

Offline kramble

  • Full Member
  • ***
  • Posts: 247
    • Github
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #56 on: October 14, 2013, 07:42:14 PM »
You just mentioned the head of engineers for our partner company. If your good at using google, maybe you can figure out who they are, and what kind of expertise they have. That's not a residential address that's our accountancy firm. You do know registered address and trading address are two different things. Anything else?

Yup, I did use google. Your accountants office.

Anyway this is getting off topic, so I'll await further details with breath abated (especially the process node and ASIC technology ... full custom, std cell, hardcopy?).

Offline AlphaTechnology

  • Jr. Member
  • **
  • Posts: 42
    • Alpha Technology
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #57 on: October 14, 2013, 07:54:16 PM »
Our office, we are also accountants. And that picture is a few years old, i wish Google would update to our new signs  :) And one of our partner's http://www.indiainfoline.com/Research/LeaderSpeak/Amit-Sinha-MD-and-CEO-Dexcel-Electronics-Designs/43387883.

Official announcements coming soon, with more relevant information.
« Last Edit: October 14, 2013, 07:58:23 PM by AlphaTechnology »

Offline kramble

  • Full Member
  • ***
  • Posts: 247
    • Github
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #58 on: October 14, 2013, 08:00:42 PM »
Our office, we are also accountants. And that picture is a few years old, i wish Google would update to our new signs  :) And one of our partner's http://www.indiainfoline.com/Research/LeaderSpeak/Amit-Sinha-MD-and-CEO-Dexcel-Electronics-Designs/43387883.

Many thanks. There is too much cloak-and-dagger in this business. Its good to see some honest disclosure for a change (not just yourself, the other players also keep their details a little too secretive for my liking).

And I was rather tickled to see the name of the accountancy firm (your family name perhaps?). Good luck in your studies (I assume you're at Manchester University, they have a very good EE department).

Offline AlphaTechnology

  • Jr. Member
  • **
  • Posts: 42
    • Alpha Technology
Re: Alpha Technology Scrypt ASIC Miner Development
« Reply #59 on: October 14, 2013, 08:17:24 PM »
A few people are involved from our end.

I love how these forums go personal.
« Last Edit: January 29, 2014, 08:16:53 PM by AlphaTechnology »