KnC's new Solar chip to obsolete all current ASICs - an end to mining diversity?

[jermwerty]

So I know KnC isn't a popular vendor around here but eventually they do tend ship a product that was somewhat close to advertised, even if late.  

If one thing is consistent about the bitcoin mining landscape is that it is always changing!  I remember sitting around with GPUs and thinking I would be soo screwed when ASICs arrived.  While that was partially true, there is still money to be made mining for those of us with cheap power and current gen ASICs ... that is until this announcement becomes real:

https://www.kncminer.com/news/news-118


Avalons started out at 10W/GH.
BFL was half that.
Antminer S1s hit 2W/GH (undervolted 1.1W/GH)
S3s are around 0.8W/GH

And while some chips are hitting around 0.6W/GH today, current generation ASICs are still within a factor of two of the power specs the S1 could achieve a year ago.

BUT with the KnC announcement they are targeting 0.07W/GH with this new 5000 core FinFET technology!

This is a big deal.  Essentially when KnC releases this chip, if it works and even hits half their estimated power efficiency, will obsolete all other ASICs instantly.

Now I know "KnC will be late" ... but IMNSHO after mining for years and "surviving" the GPU>ASIC transition, I think KnC just put the entire mining world on notice.  Let's just hope this triggers a "cat is out of the bag" moment and some of the better, more inexpensive vendors figure out how to compete.  Otherwise we might be looking at a world of Bitcoin:  brought to you by KnC!

[1715229539]

1715229539

[1715229539]

1715229539

[seriouscoin]

So I know KnC isn't a popular vendor around here but eventually they do tend ship a product that was somewhat close to advertised, even if late.  

If one thing is consistent about the bitcoin mining landscape is that it is always changing!  I remember sitting around with GPUs and thinking I would be soo screwed when ASICs arrived.  While that was partially true, there is still money to be made mining for those of us with cheap power and current gen ASICs ... that is until this announcement becomes real:

https://www.kncminer.com/news/news-118


Avalons started out at 10W/GH.
BFL was half that.
Antminer S1s hit 2W/GH (undervolted 1.1W/GH)
S3s are around 0.8W/GH

And while some chips are hitting around 0.6W/GH today, current generation ASICs are still within a factor of two of the power specs the S1 could achieve a year ago.

BUT with the KnC announcement they are targeting 0.07W/GH with this new 5000 core FinFET technology!

This is a big deal.  Essentially when KnC releases this chip, if it works and even hits half their estimated power efficiency, will obsolete all other ASICs instantly.

Now I know "KnC will be late" ... but IMNSHO after mining for years and "surviving" the GPU>ASIC transition, I think KnC just put the entire mining world on notice.  Let's just hope this triggers a "cat is out of the bag" moment and some of the better, more inexpensive vendors figure out how to compete.  Otherwise we might be looking at a world of Bitcoin:  brought to you by KnC!

If you believe that shit, i have a golden bridge to sell to you. PM me and we can discuss to close the deal

[RoadStress]

BUT with the KnC announcement they are targeting 0.07W/GH with this new 5000 core FinFET technology!

That is at chip level. At system level it will be x2-x3 higher.

[elduderino]

No offense, but I have to agree...

So these guys were the 1st to 20nm (and I don't know the details, but it was probably the hard masking method that Intel, etc simply doesn't like employing because it's more expensive than traditional masking). Doesn't mean they're going to achieve some power efficiency breakthrough simply by moving to a new (what seems to be an incremental) process? 28 nm -> 20 nm -> 16 nm?

Not likely.

[FattyMcButterpants]

BUT with the KnC announcement they are targeting 0.07W/GH with this new 5000 core FinFET technology!

That is at chip level. At system level it will be x2-x3 higher.

Even .21W/GH is significantly more efficient then the most efficient miners available today.

I don't think these will make other ASICs obsolete instantly. It would just make them less profitable.   

[brontosaurus]

So I know KnC isn't a popular vendor around here but eventually they do tend ship a product that was somewhat close to advertised, even if late.  

If one thing is consistent about the bitcoin mining landscape is that it is always changing!  I remember sitting around with GPUs and thinking I would be soo screwed when ASICs arrived.  While that was partially true, there is still money to be made mining for those of us with cheap power and current gen ASICs ... that is until this announcement becomes real:

https://www.kncminer.com/news/news-118


Avalons started out at 10W/GH.
BFL was half that.
Antminer S1s hit 2W/GH (undervolted 1.1W/GH)
S3s are around 0.8W/GH

And while some chips are hitting around 0.6W/GH today, current generation ASICs are still within a factor of two of the power specs the S1 could achieve a year ago.

BUT with the KnC announcement they are targeting 0.07W/GH with this new 5000 core FinFET technology!

This is a big deal.  Essentially when KnC releases this chip, if it works and even hits half their estimated power efficiency, will obsolete all other ASICs instantly.

Now I know "KnC will be late" ... but IMNSHO after mining for years and "surviving" the GPU>ASIC transition, I think KnC just put the entire mining world on notice.  Let's just hope this triggers a "cat is out of the bag" moment and some of the better, more inexpensive vendors figure out how to compete.  Otherwise we might be looking at a world of Bitcoin:  brought to you by KnC!

Their figure of 0.07W/ (Gh/sec) is pure fantasy, this if simply a trick to try to put doubt into their competition's minds. Look at it this way, they had plenty of time to work some magic with their Neptune machine on 20nm, yet it's power consumption is higher than most 28nm implementations. FinFet is not some magical process with a quantum leap in terms of performance, (about 50% lower power than 28nm), so unless they've done a very clever full custom design - which I sincerely doubt they have the ability to do - then the 0.07 is nonsense.

There are custom devices that exist in 28nm that can easily beat this figure by a significant margin, so I think they may well be spending a lot of money on something that is already way behind the curve.

[SMB-2525]

There are custom devices that exist in 28nm that can easily beat this figure by a significant margin, so I think they may well be spending a lot of money on something that is already way behind the curve.

Or not. How are they going to fund the development of this new generation of ASIC? I think costs increase with each decrease in spacing. With a halving coming in 2016, investment in breakthrough mining technology is problematic.

[brontosaurus]

There are custom devices that exist in 28nm that can easily beat this figure by a significant margin, so I think they may well be spending a lot of money on something that is already way behind the curve.

Or not. How are they going to fund the development of this new generation of ASIC? I think costs increase with each decrease in spacing. With a halving coming in 2016, investment in breakthrough mining technology is problematic.

Good point, and that's why jumping into a largely unproven technology is not the thing to do. All the industry pundits pretty much agree that it'll be 5 years at least before 16nm actually becomes cost effective against 28nm, except for those few applications where speed and density are essential. Bitcoin doesn't really fall into this because of the energy issue. At the moment, a 28nm mask set costs around $1.7 million, for a 16nm one it's about $7.5-8.5 million unless you're a really big customer like Qualcomm or Apple, relative to their quantities Bitcoin asic volumes are tiny.

[Grix]

And Spondoolies is targeting 0.05W/Ghps for their 2015 models, and the rest won't be far behind either. It won't kill diversity for sure, KNC is not the only one with chips like that, if they even make it.

[Sine(X)]

And Spondoolies is targeting 0.05W/Ghps for their 2015 models, and the rest won't be far behind either. It won't kill diversity for sure, KNC is not the only one with chips like that, if they even make it.

They told about 2H15, perhaps even 4Q15.. And what about 1H15, what they will do? BitFury and AM announced new chips 1Q15 or even earlier.

[Grix]

And Spondoolies is targeting 0.05W/Ghps for their 2015 models, and the rest won't be far behind either. It won't kill diversity for sure, KNC is not the only one with chips like that, if they even make it.

They told about 2H15, perhaps even 4Q15.. And what about 1H15, what they will do? BitFury and AM announced new chips 1Q15 or even earlier.

They have said that they are planning to release new 28nm chips with ambitious goal specs in Q1. The 2H is likely their 16nm chips

[seriouscoin]

There are custom devices that exist in 28nm that can easily beat this figure by a significant margin, so I think they may well be spending a lot of money on something that is already way behind the curve.

Or not. How are they going to fund the development of this new generation of ASIC? I think costs increase with each decrease in spacing. With a halving coming in 2016, investment in breakthrough mining technology is problematic.

Preorder scheme ...duh!

Remember, there are suckers born every second milliseconds

[ezeminer]

Just my opinion, either you expect too much from the first batch from a company, i.e. bitmain, or you are falling for advertising rather than a final shipped product, i.e. butterfly labs

[dwyer17]

I will never pre-order again. I don't think I'll ever buy 1st batch again either. 2nd batch is where I'm at.

[kashish948]

Mining isnt for the general public anymore. KnC will mine with your miners first and then sell it to you when its unprofitable.

[CHAOSiTEC]

Mining isnt for the general public anymore. KnC will mine with your miners first and then sell it to you when its unprofitable.

the case against BFL has proven to all manufactures, that it is considered illegal.. so i dont think there going to do so.

[MCHouston]

If KnC can find the technology to make these really efficient chips, so can all the other vendors.  They will all get around to it and keep leap frogging each other to make better chips so they can sell more product.

[xstr8guy]

Mining isnt for the general public anymore. KnC will mine with your miners first and then sell it to you when its unprofitable.

the case against BFL has proven to all manufactures, that it is considered illegal.. so i dont think there going to do so.

You need to educate yourself on the FTC's case against BFL and what that means for other manufacturers.

There are currently no criminal charges against BFL. They are being investigated by the U.S. Federal Trade Commission for fraud involving the false promises and late delivery of mining equipment to consumers.

Nothing involving the FTC vs BFL case will prevent manufacturers in other countries (such as Sweden) from doing whatever the fuck they want with mining gear before shipping to customers. KNC is currently fucking over thousands of customers by late-shipping dangerous faulty gear, not delivering promised bonus miners and generally ignoring every other conceivable good business practice. They are clearly not threatened by any legal action against BFL.

Why would you expect any other behavior from KNC with future generations of their products?!