7nm miner thread

[promojo]

7nm chips will be coming out in early 2018 (or so they say).   Let us use this as the placeholder for discussions.    

One company has talked about it lightly in their presser - Canaan.   I assume they will be restrictive and not disclose much but we should watch for news coming from TSMC.  

With that said... Who is going to continue to ramp up operations with Avalon or S9 given these are coming out?   I would figure units like that go obsolete within the first release of 7nm.

[1714689980]

1714689980

[1714689980]

1714689980

[1714689980]

1714689980

[NotFuzzyWarm]

Should be amusing to read what the masses of folks here folks with no ties to the Semiconductor Industry come up with for their Speculative dreaming thoughts. 

[Flying Hellfish]

Should be amusing to read what the masses of folks here folks with no ties to the Semiconductor Industry come up with for their Speculative dreaming thoughts. 

300ths at 500W, shipping Nov 2017! 

[fanatic26]

I like that you jumped right to 7nm and ignored the 10nm process when the mining companies are still on 14/16nm. This thread might be useful around 2020....

[wmabern]

Should be amusing to read what the masses of folks here folks with no ties to the Semiconductor Industry come up with for their Speculative dreaming thoughts.  

300ths at 500W, shipping Nov 2017!  

I will be getting three of these. Sounds great!

Make that four. I've always to have over 1ph of my own gear! 

[NotFuzzyWarm]

More to the point of the OP asking

With that said... Who is going to continue to ramp up operations with Avalon or S9 given these are coming out?   I would figure units like that go obsolete within the first release of 7nm.

Current gen obsolete soon?

I still have 1 UC/UV s7 running and only last month retired the last 2 full speed s7's. Even now they are still marginally profitable for me but I used their power quota to upgrade with Avalons. Given that I got my 1st s9 early in 2016 and that they will still be uber profitable for at least another 1-2 years there is simply no need to upgrade soon.

Now once the next lower nodes are used for mining chips - and that will be long after the main users start getting Production yields - sure. Then it maybe/probably will be economically prudent to always buy the best/latest available at that time but that will not be anytime 'soon' and definitely does not mean an existing farm is suddenly unprofitable when the next-gens finally show up in the wild.

So, in the meantime IF you are going to mine then get off yer duff and start now! Start growing your farm so it can finance future expansion/upgrades in the future.

[VRobb]

Yo NotFuzzy, shhhhh... don't be giving out sound advice. The op, and all wide-eyed naive wannabe god's gift to mining, should totally wait for the 7nm node to hit the mining hardware segment. Hell, you'd be a fool to buy in at 7nm, really, when the 4nm node will be right behind it and just think, at least 0.001J/G or 1W per THs.
You're a complete idiot for buying any current gear with THAT coming down the pike any day now!
No, really!

[NotFuzzyWarm]

Yo NotFuzzy, shhhhh... don't be giving out sound advice. The op, and all wide-eyed naive wannabe god's gift to mining, should totally wait for the 7nm node to hit the mining hardware segment. Hell, you'd be a fool to buy in at 7nm, really, when the 4nm node will be right behind it and just think, at least 0.001J/G or 1W per THs.
You're a complete idiot for buying any current gear with THAT coming down the pike any day now!
No, really!

Well sometimes that bothersome nice bit of me comes out... Usually a sign that it's time to medicate...
And don't forget that that as Foxminers claims with their 90THs Wonder miner (also using 28nm chips), "The chips are Endothermic -- instead of producing heat they actually absorb heat from their surroundings" ...

So I guess they actually cool the room and the heat majikly is transported to another dimension right? With that technology most definitely need to hold off buying today's crap...

[promojo]

You guys are all too funny.  But on a serious note; let's keep it to 7nm speculation and what it could do to the 16nm ASICs.

[fanatic26]

You guys are all too funny.  But on a serious note; let's keep it to 7nm speculation and what it could do to the 16nm ASICs.

You must have missed the point of the sarcasm dripping from every post in this thread. There is no 7nm speculation. It is so far away from being a reality there is no point in even discussing it.

[NotFuzzyWarm]

You guys are all too funny.  But on a serious note; let's keep it to 7nm speculation and what it could do to the 16nm ASICs.

You must have missed the point of the sarcasm dripping from every post in this thread. There is no 7nm speculation. It is so far away from being a reality there is no point in even discussing it.

Ja.
Case in point: This is the EUV light source used to produce the test wafers that have been ran to start characterizing how the gates actually perform vs models ASML EUV Stepper
Note the throughput given in the article: 100 wafers per hour. That was only very recently reached after almost 20 years of R&D on the light source and 100/hr is their bare minimum for the process to viable even for top-end priced chips.

From Oct 2016 about GloFo and their stepper (same one) GloFo 7nm EUV Progress

Like I said earlier: In another year or 2 if there is an economic  need for them and 10nm and lower node mining ASICs are finally produced, of course we'll buy them and of course they will start to displace the current 16/14nm node miners. Until that rather distant future (in Tech Time) is rather pointless to discuss what eventually happens to all devices as its technology evolves.

[VRobb]

Here's just the light source for the EUV - my buddy works for Cymer and installs these around the world, lucky sot!

https://www.cymer.com/euv-source/hvm-i   (crap, I see that link is dead, but look around and you'll find it...!)

It all runs under HiVac, and is a couple meters tall...!  Damn that looks like fun to me!!!

[NotFuzzyWarm]

Here's just the light source for the EUV - my buddy works for Cymer and installs these around the world, lucky sot!

https://www.cymer.com/euv-source/hvm-i   (crap, I see that link is dead, but look around and you'll find it...!)

It all runs under HiVac, and is a couple meters tall...!  Damn that looks like fun to me!!!

Good site, bookmarked to my suppliers list. From there, Breakdown on how EUV is generated
Wish I could post some pics from my toy, er, testbed at work where I'm researching the next drop in via sizes for chips... Uber-fine patterning is all good and dandy but there are also discrete layers that need to be connected to produce working chips. Bleeding-edge for those vias which are laser drilled holes then plated to conduct signals/power are currently in the low 10's of micron diameters. With the rise of 3D chip architectures our customers are demanding the ability to pack more connection points into the same or smaller areas so to that end I'm already slashing holes sizes by at least 3x. Now as with the EUV source it is more a matter of getting more average power so the process is fast enough to support consumer electronics production volumes...

[QuintLeo]

I doubt we'll see 10 nm (TSMC) or 7nm (GF) available to miner makers before 2019 or VERY VERY LATE 2018.

 The first experimental chips might show up in early 2018 or very late 2017, but there won't be capacity to spare from the BIG long-term contract customers for months after that.


 With that said - I'd predict the S11 will be around 35 TH at around 1300 watts (at the wall with the AP3++) and available in Febuary 2019 at the soonest.

[promojo]

I doubt we'll see 10 nm (TSMC) or 7nm (GF) available to miner makers before 2019 or VERY VERY LATE 2018.

 The first experimental chips might show up in early 2018 or very late 2017, but there won't be capacity to spare from the BIG long-term contract customers for months after that.


 With that said - I'd predict the S11 will be around 35 TH at around 1300 watts (at the wall with the AP3++) and available in Febuary 2019 at the soonest.

That sounds realistic.   I would be surprised to see it in mid-late 2018 to be honest.   BUT....  I have been told to watch closely @ TSMC and Canaan has made it clear their units are reduced in cost to prepare for these chips.   Are they going to simply drop the price further come 2018 when 16nm becomes more affordable due to quantity and build-out/production?

[GMPoison]

I doubt we'll see 10 nm (TSMC) or 7nm (GF) available to miner makers before 2019 or VERY VERY LATE 2018.

 The first experimental chips might show up in early 2018 or very late 2017, but there won't be capacity to spare from the BIG long-term contract customers for months after that.


 With that said - I'd predict the S11 will be around 35 TH at around 1300 watts (at the wall with the AP3++) and available in Febuary 2019 at the soonest.

That sounds realistic.   I would be surprised to see it in mid-late 2018 to be honest.   BUT....  I have been told to watch closely @ TSMC and Canaan has made it clear their units are reduced in cost to prepare for these chips.   Are they going to simply drop the price further come 2018 when 16nm becomes more affordable due to quantity and build-out/production?

We've got miners right now generating hundreds of dollars in profits per month, I don't think too many of us are concerned with when 7nm is coming out to be honest.

[promojo]

BREAKING - https://qz.com/1071926/japans-gmo-internet-group-plans-a-300-million-investment-in-bitcoin-mining/ - GMO Internet in Japan plans to develop and roll out 300M project creating their own 7nm chips in early 2018.  

The ones at risk are the ones still trying to recover their investment because they will be four times less productive,” Guiterrez says. He expects most major players will be able to upgrade as the technology for 7 nm chips becomes generally available to the foundries who do much of the work of fabricating the chips. “The other [mining chip makers] will surely follow and create their own 7 nm chips if they are not already doing it,” he says. “As [chip fabricators] get the new technology, everybody can access it.”

[philipma1957]

I do not know the chip industry but I know making money

and  here is the deal :

 pretend I am Bitmain.

I run hash nest  I make un real crazy good money with hash nest
look at the s-7 line for them

so  if they sell the s-7  by the gh and charge 35% of what it earns  you make 65% correct.

so you say fuck yeah no mess no fuss and buy it.

no auditor no regulation  but they pay you don't care.

So how about  they have zero s-7's running put them all in cold storage and use s-9's

the s-9 costs  only 16%  of the coins you  earn

huge mother fucking profit quasi legal but you can bet your left nut on it as a fact.

so they have zero incentive to build new miners  as it would sabotage their s-7 hash-nest  business

you won't see new gear from them until  the s-7  becomes a loser.

but here is the deal they have 3-4 cent power  they charge 8 cent if you use hash nest then they say that

8- 3 or 4 cents is there profit


 so  if the s-7  is really all s-9's  that 3-4 cent power cost  is effectively ½ to  1.5-2 cent 

power so if the s-7 shifts from 35% to 90%  and people dump it.  they will announce a power drop from 8 to 6.5 cents s-7 gets another extension.

they also can play games with bcc vs btc so the financial need for 7nm chips is far far far away.

and if they do  need them  they have enough cash that they will get to market ahead of any asic miner builder.

they would then wait to bring to market for us a long time.

beastly monster chips at the 7nm level may not come for 3 to 5 years as a home / farm miner

unless intel,amd,apple,nvidia decide to build an asic miner



35%   for s-7



16% for s-9




3% for L-3

[NotFuzzyWarm]

@ promojo
In addition to ^^ That is pretty much word-for-word what Bitfury said in mid-2015 about their 16nm chip and the Peta farm they were building that was going to be filled with them. Just change 'starting in' to be late Dec. 2015. We saw how well that went for all players including TSMC's biggest customers and THAT was using a node size already in pre-priduction status. In short is pure PR bullshit strictly for Investors consumption.

As of April this year there were a couple more than 10 EUV stepper systems on the planet (Cymer/ASML who make them won't give an exact number and just say "more than 10"). Right now even the most recent ones can barely produce 100 wafers/hr. Even if one is in Japan the tech is still far from doing more than characterizing results (defining models vs real-world) and building simple test structures and circuits. Not even Intel and IBM have eported going past that stage.

As for the $300million: If you think that will buy them time/space in line for 7nm tech forget it. Intel, IBM, Apple, Samsung and others have invested BILLIONS with their Foundry partners into the ever-smaller nodes sizes and they will be 1st in line for months after the tech is viable enough to be called Production-ready.

The last sentence in 2nd to last paragraph is the one single real non-puffery Truth statement in the entire PR:

“As [chip fabricators] get the new technology, everybody can access it.”

Emphasis mine...

[QuintLeo]

BREAKING - https://qz.com/1071926/japans-gmo-internet-group-plans-a-300-million-investment-in-bitcoin-mining/ - GMO Internet in Japan plans to develop and roll out 300M project creating their own 7nm chips in early 2018.  

The ones at risk are the ones still trying to recover their investment because they will be four times less productive,” Guiterrez says. He expects most major players will be able to upgrade as the technology for 7 nm chips becomes generally available to the foundries who do much of the work of fabricating the chips. “The other [mining chip makers] will surely follow and create their own 7 nm chips if they are not already doing it,” he says. “As [chip fabricators] get the new technology, everybody can access it.”

 They can plan all they want - but if the chipmaking capacity does not EXIST their plans aren't going to happen 'till it does - and established players are more likely to get access to that capacity first.
 I also question their stated plan to locate in Northern Europe, especially after KNC died with THEIR farm near a Northern Europe hydro plant....