CleverHash asic mining x11 to x15. Is this even possible ?

[djm34]

Hello,

I think we have all seen the thread about the future cleverhash asic to mine x11, x13 and x15 (all with the same asic) doing 300MH/s (or 150MH/s single core (?)) delivery one year from now
https://bitcointalk.org/index.php?topic=826541.0.

My question is simple: Is this even possible ? and how would work such asic.

I mean from my understanding of asic, an asic is designed for one task only (and between x11 and x15 there is 4 different algos).
How would that be possible ?
What could be the speed on individual algo such as groestl simd etc ? (at some point make me wonder why they don't release an asic doing every algo in the chain   they would own every coin  )

It would be good to have some answer from expert in asic/fpga field (cleverhash people are also welcome, may-be they will answer...)

Thanks

 (each time I ask this question on their thread it gets deleted... because they feel insulted or they answer with "we are in contact we best engineer in the field".... and never answer anything at all)

[Cleverhash]

It is very real.

Open-silicone, created the "gold-strike" chip, one of the first Bitcoin mining asic processors. Read about their team here, Link 1 and here, link 2 also here, link 3 and read more about "Michael Kavcak" and us here, link 4

Your posts were removed because of your trolling comments that were not only directed at us but also at our partners and other members of the community! You also quoted inaccurate misleading information, hence the removal. All your questions were answered.


Here is a quote from the Darkcoins wiki.

At this point of time (March 2014), X11 is fully ASIC-resistant/ASIC-immune as there are no ASIC implementations for it. The advanced sophistication of X11 (11 hashes combined instead of 1), will require a more complex ASIC design compared to those used for single-hash PoW. Even if ASICs are eventually deployed at some future point, it is possible that these will not be able to achieve the same degree of acceleration that was achieved in single-hashing algorithms like SHA256 as ASICs excel in paralleling workloads of decreased complexity. On the other hand, the 11 hashes comprising the X11 algorithm can theoretically be programmed very efficiently on a hardware level (FPGAs/ASICs) although the slowest hash is the one which will ultimately determine the highest speed of any hardware implementation.

Evan Duffield, the creator of Darkcoin and X11 chained-hash, has wrote on several occasions that X11 was integrated into Darkcoin not with the intention to prevent ASIC manufacturers from creating ASICs for X11 in the future, but rather to provide a similar migratory path that Bitcoin had (CPUs, GPUs, ASICs). He expects that eventually, as Darkcoin grows in market capitalization, and ASIC investments becomes profitable, ASICs will be developed.

Source: X11 ASIC resistance

[djm34]

It is very real.

Open-silicone, created the "gold-strike" chip, one of the first Bitcoin mining asic processors. Read about their team here, Link 1 and here, link 2 also here, link 3 and read more about "Michael Kavcak" and us here, link 4

Your posts were removed because of your trolling comments that were not only directed at us but also at our partners and other members of the community! You also quoted inaccurate misleading information, hence the removal. All your questions were answered.

I am just asking for questions and technical details about how you can build such chip, and answering by "we discussed with best engineer in the field" isn't a satisfactory answer (nor that they already developped bitcoin asic... which is only one algo)

Also bitcoin algo sha256 is only one algo, here what you are claiming is to be able to make an asic which run 15 algos (well why not... but a few technical details wouldn't hurt), and also how you can run x11,x13 as well as X15 all that from the same chip.

I am sorry that you feel insulted when people ask for technical detail... but I think it is a legitimate question regarding that you asking for a lot of money one year before estimated delivery (and even before R&D).  

[sp_]

VHDL sourcecode is available and opensource. Pretty easy to fit into a FPGA if the FPGA is big. Next step is ASIC but it cost alot of money. Not sure the X11,x13,x15 market is worth it.?

[BitmoreCoin]

Any combination of algorithms can be implemented in ASIC. Whether that will be implemented depends on profitability. I think Xn ASIC will not be realized due to the high cost of development and small market capitalization of Xn coins.

[Hippie Tech]

We went through this 14-16 months ago with the "scrypt asics are impossible" propaganda/ campaign. All the while, LTC's net hashrate continued to skyrocket.

That 50 mhash user wasn't gpu farm then.. and that 400 mhash x11 user isn't one now !

[djm34]

We went through this 14-16 months ago with the "scrypt asics are impossible" propaganda/ campaign. All the while, LTC's net hashrate continued to skyrocket.

That 50 mhash user wasn't gpu farm then.. and that 400 mhash x11 user isn't one now !

speculations not facts (nobody has ever been able to take a photo or anything to prove that and even after 6months, still nothing proving they really exist (buying shares doesn't count   because you have no idea of what you buy, and could be anything even a ponzy scheme  ... ) so as far as I am concerned they could be gpu farm with well optimized miner... and this we know it is possible...)

Also this thread is about the technical detail into building one and what kind of hashrate could be expected for one and not about speculating if they exist or not...

[Hippie Tech]

We went through this 14-16 months ago with the "scrypt asics are impossible" propaganda/ campaign. All the while, LTC's net hashrate continued to skyrocket.

That 50 mhash user wasn't gpu farm then.. and that 400 mhash x11 user isn't one now !

speculations not facts (nobody has ever been able to take a photo or anything to prove that and even after 6months, still nothing proving they really exist (buying shares doesn't count  because you have no idea of what you buy, and could be anything even a ponzy scheme  ... ) so as far as I am concerned they could be gpu farm with well optimized miner... and this we know it is possible...)

Also this thread is about the technical detail into building one and what kind of hashrate could be expected for one and not about speculating if they exist or not...

I'm not the one asking.. "Is this even possible ?" (see thread title)

and..

<snip>

Also bitcoin algo sha256 is only one algo, here what you are claiming is to be able to make an asic which run 15 algos (well why not... but a few technical details wouldn't hurt), and also how you can run x11,x13 as well as X15 all that from the same chip.

I am sorry that you feel insulted when people ask for technical detail... but I think it is a legitimate question regarding that you asking for a lot of money one year before estimated delivery (and even before R&D). 

Thus why I stepped in with a resounding.. "Yes it is !" .. hoping that people will learn from the past for a change.

SHA256 and scrypt asics both garnered millions in preorder sales thanks to the propaganda and hype created by threads very similar to this one.

[djm34]

We went through this 14-16 months ago with the "scrypt asics are impossible" propaganda/ campaign. All the while, LTC's net hashrate continued to skyrocket.

That 50 mhash user wasn't gpu farm then.. and that 400 mhash x11 user isn't one now !

speculations not facts (nobody has ever been able to take a photo or anything to prove that and even after 6months, still nothing proving they really exist (buying shares doesn't count   because you have no idea of what you buy, and could be anything even a ponzy scheme  ... ) so as far as I am concerned they could be gpu farm with well optimized miner... and this we know it is possible...)

Also this thread is about the technical detail into building one and what kind of hashrate could be expected for one and not about speculating if they exist or not...

I'm not the one asking.. "Is this even possible ?" (see thread title)

and..

<snip>

Also bitcoin algo sha256 is only one algo, here what you are claiming is to be able to make an asic which run 15 algos (well why not... but a few technical details wouldn't hurt), and also how you can run x11,x13 as well as X15 all that from the same chip.

I am sorry that you feel insulted when people ask for technical detail... but I think it is a legitimate question regarding that you asking for a lot of money one year before estimated delivery (and even before R&D).  

Thus why I stepped in with a resounding.. "Yes it is !" .. hoping that people will learn from the past for a change.

SHA256 and scrypt asics both garnered millions in preorder sales thanks to the propaganda and hype created by threads very similar to this one.

I hope cleverhash will give me a %age on their sales   (or and a free miner  )
The "is this even possible ?" refer to the planned hashrate 300MHash/s (or 150MHash/s per core) and doing x11 to x15 from a same hardware as it requires at least to have the slowest algos to run at that speed. That's why I am asking for more details and from what I heard from fpga groestl isn't near that sort of hashrate... (but doesn't mean much... fpga are slower than gpu on skein)

[Equate]

It is possible and as market of x11 coin grows gradually , we might see x11 Asic Pre-orders lol.

[Hippie Tech]

We went through this 14-16 months ago with the "scrypt asics are impossible" propaganda/ campaign. All the while, LTC's net hashrate continued to skyrocket.

That 50 mhash user wasn't gpu farm then.. and that 400 mhash x11 user isn't one now !

It is in fact possible to make an X11 ASIC, but no, that 400MH/s user IS probably a GPU farm. I know a guy with enough GPUs to pull several GH/s of X11 - of course, only with my code, but if I can do it, someone else probably can.

Several ghash is believable but 100's ? And that many of them ? I seriously doubt it.

Or maybe.. we should bring merged mining and stratum exploits into the conversation..

https://github.com/Crypto-Expert/stratum-mining/issues/313

[djm34]

actually what people seem to disregard is the link between the rise of x11 hashrate and srypt asics.
It seems obvious at least to me that many moved away their gpu farm from scrypt and started to mine any x11 crap around...

Did a few x11 coin launch, there was right away from the start several GH/s... now that might plead for x11 fpga... but the same behavior was seen on second x15 coin  (which was launched only a couple of day after the gpu algo had been released... ) and it couldn't have been fpga because of the timing (as they were not present on first x15 coin and I don't think fpga dev need to wait for the gpu code to get released before running...   )

Also same behavior of freshcoin/algo; same behavior on Nist algo.... and if you look carefully to what it is pointing you will see hash renting company (and investor/non-miner joining in through that) and not fpga.

So basically the "fpga rumor" was created, in my opinion, mostly as fud and/or to counteract against people moving away from scrypt... (probably also against the infamous thread asking ltc to change to x11 algo)
(I don't have anything against chinese altcoin supporters, but they are pretty good at fuding and making up things... (got a few things made up against me while working on talkcoin  )


ps: it doesn't mean that I don't think fpga/asic can't be developped, it just means the timing was not right (also don't forget that it also took some time to darkcoin to gain traction... )

[Hippie Tech]

We went through this 14-16 months ago with the "scrypt asics are impossible" propaganda/ campaign. All the while, LTC's net hashrate continued to skyrocket.

That 50 mhash user wasn't gpu farm then.. and that 400 mhash x11 user isn't one now !

It is in fact possible to make an X11 ASIC, but no, that 400MH/s user IS probably a GPU farm. I know a guy with enough GPUs to pull several GH/s of X11 - of course, only with my code, but if I can do it, someone else probably can.

Several ghash is believable but 100's ? And that many of them ? I seriously doubt it.

Or maybe.. we should bring merged mining and stratum exploits into the conversation..

https://github.com/Crypto-Expert/stratum-mining/issues/313

Who has hundreds of GH/s on X11? One entity?

I first became aware of this last May when .. over at the Logicoin thread..

some unknown holding more than 70% of Hashrate 

I'll guess that its one, or a few, of these giant X11 farms that seem to be getting popular.

Here are two more known stratum bugs, posted on the following page of that same thread.

some unknown holding more than 70% of Hashrate 

http://i60.tinypic.com/2mqpamx.png

i wouldn't worry about that too much. that doesn't mean it's all coming from one pool.

after a little bit of snooping, here's one of the unknown pools http://digger.pw/logi/en.

block halving rapidly approaching. nethash is through the roof.

Thats a good news and also good volume on Poloniex.

Not if the hashpower is fake.

fake hash vulnerability  #66
https://github.com/slush0/stratum-mining-proxy/issues/66

Flooding with shares below the pool target #326
https://github.com/Crypto-Expert/stratum-mining/issues/326

[senseless]

Cleverhash is probably real based on what I saw on their site. At least, their technical jargon was in order and their hashrate to power usage is about what is on par with what I expect out of a highly optimized 28nm cell based asic. HOWEVER -- their estimates are extremely generous. I would expect the device to perform at about 1/2 of the hashrate to power usage they say it will. FYI, if they were to publish the conservative figures of around 200Mh/s per 1KW no one would buy because it's not going to be profitable.

Open-Silicon has a pretty impressive first run success record. But, they are a huge and slow company. I expect the timeline to be longer than what they're suggesting. The RTL for x11 is ridiculous and x15 is even worse.

We started our RTL about a month before they (cleverhash) did and we're now done and moving on to the design phase (without pre-order money).

To the other's that are suggesting there are x11 or x15 asics in the wild -- I doubt there are any X11 or X15 ASICs in the wild. Reason being, the cheap method to produce quick chips (structured asic ala KNC, Avalon, etc) isn't cost effective for these designs as they're too big. The only way to do these designs is cell based asic which is going to be a large lump sum NRE. Unfortunately the lump sum is about 10-20% the market cap of Darkcoin. Just to give you an idea of what we're talking in terms of speed, an FPGA I was achieving 1Gh/s at 20W with on bitcoin mining will do about 50Kh/s on X11. As far as I'm aware -- publicly -- I had the fastest single FPGA mining bitcoin for a year or more until the last gen (28nm) Xilinx chips came out.

Other than that, it takes forever to put the RTL together. Imagine designing 11 or 15 different bitcoin designs and putting them all on one chip. If you thought making a bitcoin chip was tough, you haven't seen nothing yet. There are 15 different cores in their RTL. Each core is equivalent to making an entire double sha2 bitcoin mining RTL. I doubt you're going to see any of those cheap and quick chinese manuf pick up x11. They like to skip steps and cut corners to get their designs done faster. I made around $30K from my B1 avalon, I'm not complaining. But at the same time, X11/X15 is a wholly different devil.

Pro Tip: Don't pre-order anything, ever. You have better odds taking all your money to the casino and putting it on black at the roulette table. Or my personal favorite, playing hard ways at the craps table.

I'm amazed they're trying to go the pre-order route on an X15 ASIC. It's pure insanity on their part. Though they do have a good pump with that whole clevercoin thing or whatever they call it. A sucker is born every minute. 

[Amph]

well in the end asic will mine everything that is worth mining, just a matter of time, they problem is just the dedicated thing, other than that they are a gpu/cpu killer

just release new algo everytime to keep asic away from our "food"(deteriorated food  )

That actually could be real, the top one - 1GH/s on X11 is very doable with GPUs - as I said, I know someone with a farm capable of several.

for sure, just 100 290x, in germany there was a gui with 200+ 280x