Anyone demonstrated a 28nm ASIC mining bitcoin yet?

[Fuzzy]

There's a lot of discussion about 28nm ASICs out there, but I can't find any proof that shows one actually running, with a hashing output and a watt meter.

I know "proof" is subjective, but besides receiving a unit and testing it myself, a video or review by a trusted 3rd party would suffice for me.

[1714139215]

1714139215

[Bitcoinorama]

http://www.toshiba.com/taec/Catalog/Family.do?familyid=1

[dogie]

http://www.toshiba.com/taec/Catalog/Family.do?familyid=1

KNC sock puppet avoids the serious question and posts stupid link. If you wanted to show that 28nm is possible, why wouldn't you have mentioned the current and past 22nm intel chips a lot of us are using?

Ah yes, because you 10000% blindly believe in KNC.

tldr: Answer is no.

[Fuzzy]

http://www.toshiba.com/taec/Catalog/Family.do?familyid=1

Make my day! Say thanks if you found me helpful

har har

[Bitcoinorama]

http://www.toshiba.com/taec/Catalog/Family.do?familyid=1

KNC sock puppet avoids the serious question and posts stupid link. If you wanted to show that 28nm is possible, why wouldn't you have mentioned the current and past 22nm intel chips a lot of us are using?

Ah yes, because you 10000% blindly believe in KNC.

tldr: Answer is no.

Try for once not to be a dick.

[dogie]

http://www.toshiba.com/taec/Catalog/Family.do?familyid=1

KNC sock puppet avoids the serious question and posts stupid link. If you wanted to show that 28nm is possible, why wouldn't you have mentioned the current and past 22nm intel chips a lot of us are using?

Ah yes, because you 10000% blindly believe in KNC.

tldr: Answer is no.

Try for once not to be a dick.

Try once not to avoid the topic in question.

[Bitcoinorama]

http://www.toshiba.com/taec/Catalog/Family.do?familyid=1

KNC sock puppet avoids the serious question and posts stupid link. If you wanted to show that 28nm is possible, why wouldn't you have mentioned the current and past 22nm intel chips a lot of us are using?

Ah yes, because you 10000% blindly believe in KNC.

tldr: Answer is no.

Try for once not to be a dick.

Try once not to avoid the topic in question.

How did I avoid it? OP asked a non-specific Q as to whether anyone had produced a working 28nm ASIC, so I posted a link.

Turns out the OP has a sense of humour, unlike some.

[Syke]

Bitfury has the smallest demonstrated chip at 55 nm.

[lumpycustard]

None on 28nm, although Hashfast demonstrated their 28nm design running CGminer, but it was just being tested on a FPGA.  Supposedly there will be half a dozen 28nm products before the end of the year, while only 2 years ago 28nm was bleeding edge.  Makes you wonder which ones are going to fail hard... surely at least one.

[Fuzzy]

Bitfury has the smallest demonstrated chip at 55 nm.

I can't find any demonstrated hashing output and power consumption examples for Bitfury, only claims 

[Bitcoinorama]

Bitfury has the smallest demonstrated chip at 55 nm.

I can't find any demonstrated hashing output and power consumption examples for Bitfury, only claims 

https://metabank.ru/asic

[Syke]

Bitfury has the smallest demonstrated chip at 55 nm.

I can't find any demonstrated hashing output and power consumption examples for Bitfury, only claims 

Nice video: https://www.youtube.com/watch?v=kYyrNhLwrn0

[Fuzzy]

Bitfury has the smallest demonstrated chip at 55 nm.

I can't find any demonstrated hashing output and power consumption examples for Bitfury, only claims 

Nice video: https://www.youtube.com/watch?v=kYyrNhLwrn0

0.7 W/GH? on a 55nm chip? Impressive.

[Bitcoinorama]

Bitfury has the smallest demonstrated chip at 55 nm.

I can't find any demonstrated hashing output and power consumption examples for Bitfury, only claims 

Nice video: https://www.youtube.com/watch?v=kYyrNhLwrn0

0.7 W/GH? on a 55nm chip? Impressive.

Indeed, I was looking for that video, Skye beat me to it!

[DeathAndTaxes]

Bitfury has the smallest demonstrated chip at 55 nm.

I can't find any demonstrated hashing output and power consumption examples for Bitfury, only claims  

Nice video: https://www.youtube.com/watch?v=kYyrNhLwrn0

0.7 W/GH? on a 55nm chip? Impressive.

That is clocked lower than the production board.  The demo is running @ 1.2 GH/s per chip and the stock boards (H-board) are running at 1.56 GH/s.  c-scape reported specs on a third party board (with more powerful DC power supply) running at 2.5 GH/s per chip.

Demo:    1.20 GH/s @ 0.84W = 0.70 W/GH
H-Board: 1.56 GH/s @ 1.48W = 0.95 W/GH *
S-HASH: 2.50 GH/s @ 2.18W = 0.87 W/GH

* Not demoed publicly.  Power seems out of line with the other two boards.  It may be the reported wattage (380W on 400 GH/s) is conservative or is measured at the wall and includes AC PSU inefficiency. Unsure.

Some more specs (temp, wattage, efficiency) of the demo board running at various clock speeds:
https://bitcentury.io/blog/initial-testing-of-bitfury-asic

On edit: fixed math error.  division can be hard. Thanks Keefe.

[Keefe]

BFL is predicting their 28nm chips will use barely less W/GH than BitFury's 55nm chips, when they should be at least 3x better than BitFury, if their design was as good. BitFury's 28nm chips (if they plan to do that) ought to be awesome.

[Fuzzy]

BFL is predicting their 28nm chips will use barely less W/GH than BitFury's 55nm chips, when they should be at least 3x better than BitFury, if their design was as good. BitFury's 28nm chips (if they plan to do that) ought to be awesome.

Without getting into BFL's 28nm plans, it's boggling how their 65nm design is eating 4w/GH even after their epic development time, whereas this 55nm design is well under 1w/GH and seems to have come out of no where.

[Gomeler]

BFL is predicting their 28nm chips will use barely less W/GH than BitFury's 55nm chips, when they should be at least 3x better than BitFury, if their design was as good. BitFury's 28nm chips (if they plan to do that) ought to be awesome.

Without getting into BFL's 28nm plans, it's boggling how their 65nm design is eating 4w/GH even after their epic development time, whereas this 55nm design is well under 1w/GH and seems to have come out of no where.

BitFury's design didn't come out of nowhere. His FPGA design(from what I've read) was superior to any other design on the market. Reading the little information that was fed to us, BitFury's 65nm shrunk 55nm chips were designed to be a low-voltage and efficient design. They missed their hashing/chip goals but the design has been proven to display fantastic efficiency and power-consumption/throughput scaling.

A 28nm poorly designed chip could easily be worse than a well designed 55nm chip. I am excited for all of these upcoming 28nm chips but I'm really excited to see what comes out of BitFury(28nm die shrink perhaps?) and CoinTerra(due to the credentials of the team). I think this next generation of chips will be the generation that we mine on for a year+ as great designs will be limited by the manufacturing process, unlike the past where 110-180nm designs had a lot of room to scale with optical shrinks.

[DeathAndTaxes]

@Gomeler  Agreed on both points.  Design matters and we won't be moving past 28nm for a while.  

As for design there is something horribly broke in BFL design and I can't see them doing a significant redesign at this point.  Bitfury and BFL aren't the only two with vastly different efficiencies at the same scale.  Granted both of these are vaporware at this point but I find it interesting that Hashfast is reporting ~0.6W/GH (at the chip) and KNC is reporting ~2.2W/GH (at this chip) and both of them are on a 28nm process.  We will see when it gets to real silicon but if that holds it is interesting.

As for next gen.  My guess is we won't be seeing anything smaller than 28n for a long time. 20nm is expensive and there is also a troubling trend which Nvidia highlighted a while back in that the cost per transistor isn't going down much even when processes become mature so the smart money may be on optimization.    I do think there will be incremental improvements on the same process (the "tock" in Intel tick tock strategy).  I wonder how efficient 28nm chips can get (GH/mm2 and GH/J).  When the low hanging fruit is gone it comes down to who can build the better company.  All that boring stuff like finding solid suppliers, optimizing pricing so you can reduce incremental cost through larger runs, building an OEM network.  I think eventually most of the ASIC companies will just produce chips and leave the rest to partners.  I also think there will be less companies operating in 2014 either through failures or consolidation.

[Gomeler]

@Gomeler

Agreed on both points.  Design matters and we won't be moving past 28nm for a while. 

As for design there is something horribly broke in BFL design and I can't see them doing a significant redesign at this point.  Bitfury and BFL aren't the only two with vastly different efficiencies at the same scale.  Granted both of these are vaporware at this point but I find it interesting that Hashfast is reporting ~0.6W/GH (at the chip) and KNC is reporting >2W/GH (at this chip) and both of them are on a 28nm process.  We will see when it gets to real silicon but if true that is significant. 

As for next gen.  We won't be seeing anything smaller than 28nm.  22nm is cost prohibitive and you are looking at huge NRE.  By the time it becomes possible the market will be flooded.  I do think there will be incremental improvements on the same process (the "tock" in Intel tick tock strategy).   There is also a troubling trend which Nvidia highlighted a while back in that the cost per transistor isn't going down much even when processes become mature so the smart money may be on optimization. 

The 110/130nm tech is going to be obsolete very soon.  I don't mean people won't be able to mine with them I just mean nobody is going to be able to sell new units.  If AsicMiner and Avalon are going to stay in the game they will be forced to make a move to 55nm if not 28nm.  When everyone is on the same process it is going to come down to design, system optimization, larger runs to reduce incremental cost, and having a solid supply chain.  It will be interesting.   I don't think every company that is around today will still be around a year from now.

I agree, it sounds like BFL there is something wrong with BFL's chip. Perhaps it has a chunk of the chip that is unusable but they can't disable. We don't know the finer details but it might not be out of the question that they are taking their 65nm design, correcting the issues and shrinking it to 28nm. I'm not completely familiar with the process but I suspect this would involve less engineering effort especially if they aren't hand-placing the design.

I'm not so sure there will be that many incremental changes made to chips once they start to hit the process limit. I say this as I expect the margins on chips to become so slim that it would be financial ruin to invest in the NRE for another 28nm chip once you already have one. Unless of course we discover a major optimization to SHA256 which would yield sufficient sales to recover the NRE and turn a profit. I fully expect we'll hit a lower-bound on energy efficiency for a given process (let's say 0.25W/GH for sake of argument) and then it'll come down to the GH/wafer as everybody competes on maximizing profit/wafer and GH/USD.

[DeathAndTaxes]

I'm not so sure there will be that many incremental changes made to chips once they start to hit the process limit. I say this as I expect the margins on chips to become so slim that it would be financial ruin to invest in the NRE for another 28nm chip once you already have one.

I guess it all depends on how "good" existing designs are.  Taking HF just because they provided die size, hashrate and power info.

One 18x18mm die is able to do 400 GHash (nominal - more overclocked**)
   Hashing per square mm:
      18x18mm = 324mm^2
      400 GHash / 324mm^2 = 1.23 GHash/mm^2

Reported power consumption (at the chip) is 400 GH / 250 W = 1.6 GH/W.

So HF claims their silicon is good for 1.23 GH/mm^2 and 1.6 GH/W @ 28nm.  
While they are impressive compared to 65nm tech it remains to be seen how "good" those are compared to what is possible for 28nm tech.  Are they very good, or barely adequate in the grand scheme of things?   

If the run some simulations on an improved die and it shows to only be marginally better (say 1.4 GH/mm2  and 1.8 GH/W) then I agree the chip probably won't be made before moving to a smaller process. On the other if they came up with say >2.0 GH/mm^2 and >3 GH/W) it may make sense to produce a second 28nm chip.   I just wonder how much improvement is possible,  I guess once we get a couple of vendors with real 28nm silicon we should have a better idea.

[Gomeler]

I'm not so sure there will be that many incremental changes made to chips once they start to hit the process limit. I say this as I expect the margins on chips to become so slim that it would be financial ruin to invest in the NRE for another 28nm chip once you already have one.

I guess it all depends on how "good" existing designs are.  Taking HF just because they provided die size, hashrate and power info.

One 18x18mm die is able to do 400 GHash (nominal - more overclocked**)
   Hashing per square mm:
      18x18mm = 324mm^2
      400 GHash / 324mm^2 = 1.23 GHash/mm^2

Reported power consumption is 400 GH / 250 W = 1.6 GH/W.

So HF claims 1.23 GH/mm^2 and 1.6 GH/W @ 28nm.  It remains to be seen how "good" those are for 28nm.  Are they very good, or barely adequate?

If their simulations show an improved version is only marginally better (say 1.4 GH/mm2  and 1.8 GH/W) I agree it wouldn't make sense to try and squeeze out more efficient chips.  On the other hand say >2.0 GH/mm^2 and >3 GH/W) are possible.   Once we get multiple vendors with stats based on real silicon we can start to get a better idea of relative efficiency.

Just responding to say that I agree on all points. I imagine the first 28nm designs may be more safe than elegant to minimize risk, leaving performance on the table. 2014 will be exciting, 2015 maybe not so much.

[TheSwede75]

Bitfury has the smallest demonstrated chip at 55 nm.

I can't find any demonstrated hashing output and power consumption examples for Bitfury, only claims 

Obviously not been looking. Bitfury's private pool is hashing at 65TH using the 55nm as we speak.

[Fuzzy]

I'm not so sure there will be that many incremental changes made to chips once they start to hit the process limit. I say this as I expect the margins on chips to become so slim that it would be financial ruin to invest in the NRE for another 28nm chip once you already have one.

I guess it all depends on how "good" existing designs are.  Taking HF just because they provided die size, hashrate and power info.

One 18x18mm die is able to do 400 GHash (nominal - more overclocked**)
   Hashing per square mm:
      18x18mm = 324mm^2
      400 GHash / 324mm^2 = 1.23 GHash/mm^2

Reported power consumption is 400 GH / 250 W = 1.6 GH/W.

So HF claims 1.23 GH/mm^2 and 1.6 GH/W @ 28nm.  It remains to be seen how "good" those are for 28nm.  Are they very good, or barely adequate?

If their simulations show an improved version is only marginally better (say 1.4 GH/mm2  and 1.8 GH/W) I agree it wouldn't make sense to try and squeeze out more efficient chips.  On the other hand say >2.0 GH/mm^2 and >3 GH/W) are possible.   Once we get multiple vendors with stats based on real silicon we can start to get a better idea of relative efficiency.

Just responding to say that I agree on all points. I imagine the first 28nm designs may be more safe than elegant to minimize risk, leaving performance on the table. 2014 will be exciting, 2015 maybe not so much.

Maybe in terms of bitcoin mining. But the snow globe's only just started to shake as far as crypto-currency goes 

[Loredo]

I hafta say:  some of the replies on this thread contain the kind of information and informed opinions I came to this forum to read. 

Thanks to...I don't have to say; you all know if I'm referring to you.

[CoinHoarder]

I'm not so sure there will be that many incremental changes made to chips once they start to hit the process limit. I say this as I expect the margins on chips to become so slim that it would be financial ruin to invest in the NRE for another 28nm chip once you already have one.

I guess it all depends on how "good" existing designs are.  Taking HF just because they provided die size, hashrate and power info.

One 18x18mm die is able to do 400 GHash (nominal - more overclocked**)
   Hashing per square mm:
      18x18mm = 324mm^2
      400 GHash / 324mm^2 = 1.23 GHash/mm^2

Reported power consumption (at the chip) is 400 GH / 250 W = 1.6 GH/W.

So HF claims their silicon is good for 1.23 GH/mm^2 and 1.6 GH/W @ 28nm.  
While they are impressive compared to 65nm tech it remains to be seen how "good" those are compared to what is possible for 28nm tech.  Are they very good, or barely adequate in the grand scheme of things?  

If the run some simulations on an improved die and it shows to only be marginally better (say 1.4 GH/mm2  and 1.8 GH/W) then I agree the chip probably won't be made before moving to a smaller process. On the other if they came up with say >2.0 GH/mm^2 and >3 GH/W) it may make sense to produce a second 28nm chip.   I just wonder how much improvement is possible,  I guess once we get a couple of vendors with real 28nm silicon we should have a better idea.

I would venture to say that HashFast's implementation of 28nm process is not as good as it gets, and is more of a race to the market type design.

I think Cointerra will beat them in electrical efficiency judging by their resumes. "CoinTerra boasts a highly experienced engineering team of semiconductor architects and designers who have previously designed some of the world’s highest performance CPUs, GPUs and chipsets for NVIDIA, Intel, Samsung, Qualcomm and Nortel. Having worked on several generations of low-power mobile devices, our team brings tremendous experience in power efficient circuitry, design methodology and implementation to the exciting new frontier of Bitcoin mining. "

They are claiming "significantly less than a watt per Gh/s" on their 28nm product.

[Fuzzy]

I would venture to say that HashFast's implementation of 28nm process is not as good as it gets, and is more of a race to the market type design.

I think Cointerra will beat them in electrical efficiency judging by their resumes. "CoinTerra boasts a highly experienced engineering team of semiconductor architects and designers who have previously designed some of the world’s highest performance CPUs, GPUs and chipsets for NVIDIA, Intel, Samsung, Qualcomm and Nortel. Having worked on several generations of low-power mobile devices, our team brings tremendous experience in power efficient circuitry, design methodology and implementation to the exciting new frontier of Bitcoin mining. "

They are claiming "significantly less than a watt per Gh/s" on their 28nm product.

Considering we already see <0.7w/GH on 55nm parts, qualifying the claim with "significantly less" as rather arbitrary for 28nm tech. I want to see these come down to fractions