Bitfury: "16nm... sales to public start shortly"

[64dimensions]

Along with the fixed string voltage option, I'm wondering why ATX power supply manufacturers don't use a move from the old HP playbook and come out with an option "B" for their high wattage power supplies. Option B would entail having a recessed screwdriver driven adjustment of the 12V bus. There are so many user communities that could use an output from +12V to +13V+. Functionally the miner designers should off load voltage control and efficiency to the power supply types anyway.

[1715100273]

1715100273

[1715100273]

1715100273

[Mitak]

Good idea , but I think the problem is how much these PSU's will cost.
I think there are adjustable Voltage 12V PSU's , just the price tag is not ''good"

[Dobmaster]

Just when I was hoping difficulty increases would stabilize a little bit this will keep the ball rolling for a little longer if they do release this soon.

The difficult will keep on rise for at least 10% every two weeks for the next 3 months. Bitmain is still supplying the market and so will the Bitfury.

[QuintLeo]

Along with the fixed string voltage option, I'm wondering why ATX power supply manufacturers don't use a move from the old HP playbook and come out with an option "B" for their high wattage power supplies.

 Not part of the ATX spec, and would add significant cost to them.
 They would also be less efficient, as they can't be designed to maximise efficiency at a specific operating voltage.



 The whole point to a string design is to not NEED a voltage regulator on the board at all, dropping the cost substantially. The tradeoff is flexability to tailor the voltage for max output or max efficiency.


 I don't see Bitfury ever being listed on the NYSE, as they're too "secretative", they're too SMALL, and they are not a US company anyway (depository shares at MOST, similar to what some Canadian companies like Pengrowth do, but more likely on NASDAQ than NYSE).

[64dimensions]

Along with the fixed string voltage option, I'm wondering why ATX power supply manufacturers don't use a move from the old HP playbook and come out with an option "B" for their high wattage power supplies.

 Not part of the ATX spec, and would add significant cost to them.
 They would also be less efficient, as they can't be designed to maximise efficiency at a specific operating voltage.



 The whole point to a string design is to not NEED a voltage regulator on the board at all, dropping the cost substantially. The tradeoff is flexability to tailor the voltage for max output or max efficiency.


 I don't see Bitfury ever being listed on the NYSE, as they're too "secretative", they're too SMALL, and they are not a US company anyway (depository shares at MOST, similar to what some Canadian companies like Pengrowth do, but more likely on NASDAQ than NYSE).

Sorry, not quite true.

At some point in time, some ATX power supplies were built with fine adjust pots for the different output voltages:

http://hardforum.com/showthread.php?t=765175

In terms of efficiency, if there is an on board final adjust pot, this would only move the +12V output bus by a few percent. Doubtful that it would change the efficiency much. The point I'm making is that if there are designs that still have this onboard pot, this is useful info for the A6 and other string miner users. The tradeoff being that you void the warranty by opening up the case and also expose yourself to hazardous voltages.

[dogie]

I don't see Bitfury ever being listed on the NYSE, as they're too "secretative", they're too SMALL, and they are not a US company anyway (depository shares at MOST, similar to what some Canadian companies like Pengrowth do, but more likely on NASDAQ than NYSE).

If you look at the case docket in the Bitfury vs exCFO case, they have incorporations all over the place, including at least 2 in the US that I remember.

[rupy]

Would be nice with some passively cooled 16nm chips to heat the cottage in the forest. The Monarch is awesome for heating when underclocked but seen how they meltdown completely if the fan breaks (firehazard, I use two fans in push/pull and fingers crossed) and also I'm thinking 16nm is the last major efficiency improvement.

[QuintLeo]

14/16nm full custom will be the last major efficiency improvement - for a few years.

 IBM has already announced they're researching 8nm (IIRC) for the next generation process.

[64dimensions]

14/16nm full custom will be the last major efficiency improvement - for a few years.

 IBM has already announced they're researching 8nm (IIRC) for the next generation process.

So here is what is confusing. If 14/16nm is due out within a few months, this would severely shorten the useful life of the A6/S7 models. Yet the network hash seems to say that a lot of money was dumped on these very models. If true, even the big boys might have problems with ROI. One wonders if BITMAIN et al would recoup their 1385 development costs. You would think the the competitive intel of the bigs would be accurate.

Speaking of IBM, one of their old tricks used to be to pre announce their next model of large systems to forestall customers from buying their competitors hardware.

[GSpgh]

14/16nm full custom will be the last major efficiency improvement - for a few years.

 IBM has already announced they're researching 8nm (IIRC) for the next generation process.

So here is what is confusing. If 14/16nm is due out within a few months, this would severely shorten the useful life of the A6/S7 models. Yet the network hash seems to say that a lot of money was dumped on these very models. If true, even the big boys might have problems with ROI. One wonders if BITMAIN et al would recoup their 1385 development costs. You would think the the competitive intel of the bigs would be accurate.

Speaking of IBM, one of their old tricks used to be to pre announce their next model of large systems to forestall customers from buying their competitors hardware.

of course Bitmain will recoup their costs

their miners are selling as fast as they can make them

their customers might not be as lucky though

not sure about Avalon they aren't interested in selling in small quantities and their pricing sucks

[Mitak]

But Avalon is better tech anytime compared to Bitmain.
Also SW is open source By GPL, something you can not say about BITMAINtech

[GSpgh]

But Avalon is better tech anytime compared to Bitmain.
Also SW is open source By GPL, something you can not say about BITMAINtech

better tech than only works at 75°C?

in any case it's not only the tech that makes a successful manufacturer... marketing has a lot to do with it and Bitmain is way ahead there

but this is BitFury thread who don't even have the "tech" yet just threaten everyone with it... not a fair fihgt so far

[sidehack]

If by "don't have the tech yet" you mean "haven't completed mass production of the now proved-with-engineering-samples tech yet", then sure. But unless they're outright lying on this press release, they have the tech. Silicon in hand.

I would venture to say Bitmain's already more than paid off their dev costs for the BM1385. They've been selling S7 like made for what, three months or so? Most of those were priced well above $10 per ASIC, and if they're anything like BM1384 they'd still profit from chips at below $3 per ASIC. I know there are additional costs to manufacture, like all the other complex high-dollar components on their boards (oh wait, minimal BOM design), but still. Now they're selling at about $1100 a pop, and I'm pretty sure the unit cost hasn't decreased, so you know they made at least $700 per when they sold at $1800 each. I'd guess it was well over $1000 margin, which it only takes 1000 units to make back a million bucks. Heck my tiny facility has more than a dozen running in it. They've shipped 8 batches, how many to a batch?

And Bitmain already said when the BM1385 rolled out that they were basically sitting on a full-custom 16nm design, just waiting for the right (which is to say, most profitable) time to deploy it. They know what they're doing.

[NotFuzzyWarm]

14/16nm full custom will be the last major efficiency improvement - for a few years.

 IBM has already announced they're researching 8nm (IIRC) for the next generation process.

Agreed. After several years of announcing 14/16nm is 'soon to be here' the foundries like TSMC, GloFo, Samsung et al are only now finally nailing down the production processes for 14/16nm chips with reasonable yields.

Still have massive dead chip rates vs the larger nodes but getting 30-40% viable chips per run ain't bad. The current scrap rates are very acceptable for companies (cell phone/PC chip makers) that can afford it but still relate to barely acceptable for other chips users. Still, give it time and yield rates will continue to climb.

As for the 10-8nm nodes...   Not going to see that fore at least 3-5 more years assuming they get the EUV light sources to be more powerful and stable. The current way being used (zapping droplets of tin with >30kw lasers and collecting the light produced) is working but IMHO not the final solution. re http://optics.org/news/6/6/31 and http://www.photonics.com/Article.aspx?AID=56802

[Mitak]

For sure Bitmain know what they are doing I just personally do not like them .
I like Avalon better, they do not behave as most of Chinese "eat all" companies, at least until now
The greatest advantage of Bitfury is that they know exactly what to do and how and not last they are not Chinese

[GSpgh]

If by "don't have the tech yet" you mean "haven't completed mass production of the now proved-with-engineering-samples tech yet", then sure. But unless they're outright lying on this press release, they have the tech. Silicon in hand.

well "technically" LOL you're right I guess

[Finksy]

If by "don't have the tech yet" you mean "haven't completed mass production of the now proved-with-engineering-samples tech yet", then sure. But unless they're outright lying on this press release, they have the tech. Silicon in hand.

Lying or stretching the truth?  "Tape-out" was supposedly completed in August-September, http://www.coindesk.com/bitfury-completion-16nm-bitcoin-mining-asic/ but what, the chips were sitting in storage until now when they had the time to do a press release?  I'm with the others, I'll believe it when I see it.

[sidehack]

today announced mass-production of its full-custom design 16nm Application Specific Integrated Circuit (ASIC) Chip

On average, the measured power efficiency of tested engineering samples of the new BitFury Chip ranges from 0.055 joules per gigahash to 0.07 joules per gigahash

Doesn't sound like much room for spin there. What they said is pretty clear, so it's either truth or lies. Announcing "mass production" is fairly grey, as there are who knows how many steps required and they might just be on step 1, but they outright stated in plain terms that they had tested engineering samples so the design has been verified in the real world. Maybe the announced tapeout was for these samples? That would have given them chips in about November, and then a month testing the crap out of 'em I guess before moving on to full-scale production? Not impossible. I sure hope they're not lying about having samples in hand because I also hope they're not lying about making them available because I really want to play with these chips.

[rammy2k2]

my opinion is they already mine on those "new" chips, profits were made, and now they release for public too

[QuintLeo]

If by "don't have the tech yet" you mean "haven't completed mass production of the now proved-with-engineering-samples tech yet", then sure. But unless they're outright lying on this press release, they have the tech. Silicon in hand.

Lying or stretching the truth?  "Tape-out" was supposedly completed in August-September, http://www.coindesk.com/bitfury-completion-16nm-bitcoin-mining-asic/ but what, the chips were sitting in storage until now when they had the time to do a press release?  I'm with the others, I'll believe it when I see it.

 Tapeout is normally a step MONTHS in advance of producing actual sellable/manufacture with chips.
 Sometimes, if the design wasn't done right, tape-out is just "oops, we designed these but the design don't WORK back to the drawing board to figure out how to FIX what's not working right", though that's not common with modern design tools. More commonly, the design works but doesn't meet the intended specs at which point you have to try to figure out how to fix the design to meet target specs, or you decide to live with the specs achieved if they're close enough to target.

 Even in cases where the tape-out prototype chips meet the target specs, actually ramping up to production is still usually a "months to go" thing.


 It's not hard to picture Bitfury having working chips NOW though, all of their announcements have been a few months ahead of the corresponding Innosilicon ones, and I'm expecting full production out of Innosilicon 1Q next year timeframe.