Pages:
Author

Topic: Bitfury: "16nm... sales to public start shortly" - page 49. (Read 108588 times)

legendary
Activity: 1666
Merit: 1185
dogiecoin.com
Why do you think they are selling their Golden Geese?  

Because it makes more money. The more of your own equipment you put on the network, the lower your returns.

The first miner competes 0% against yourself. After 10% of the network, your miners compete against themselves 10% of the time [shitty metric but its a visualization]. At 50% of the network, the next miner turned on competes half against the rest of the network and half against your own miners, significantly reducing its potential income. But if you distribute hashing power to everyone, you can keep your network percentage lower (and so miner returns better), while also recouping 100% of the theoretical mining earnings by selling the rights to those earnings.
legendary
Activity: 4256
Merit: 8551
'The right to privacy matters'
Philip: Wouldn't is be possible with immersion cooling with 3M novec 7000 in a 2 phase system?



okay we know that immersion lets you make the chips more dense.

but no matter what the tank of coolant is sooner of later it reaches its max in heat storage.  that is why you have fans to toss that heat away.

I see 16 fans  on the container  

those fans are 3 to 4 feet wide  lets say 4

here are the fans you need

http://trianglefans.com/jet-upblast/

http://trianglefans.com/jd/

they have some 42 inch models that will move 30,000 cfm

 I guess 16 of them would work as that is about 500,000 cfm

the avalon6 uses 200cfm to move heat from 1kwatt  and 2000 avalon6's use 2 mega watts so

2000 x 200 = 400,000 cfm  So I guess the air could be moved

legendary
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
I was told last spring that BM1384 could be had by the batch for $2.50 per chip. Assuming they still have some profit on that, and if BM1385 cost about the same to make, a 45-chip S7 would have under $300 worth of ASIC. Not sure what everything else would cost, but I was betting three months ago that they could sell for $600 and still profit (on materials, not necessarily chip dev costs).
sr. member
Activity: 441
Merit: 250

Like everything that Bitfury announces you have to read very carefully, I refer to: "We understand it will be nearly impossible for any older technology to compete .......". The 'nearly' is the giveaway, despite their strange announcement about 'every transistor on the new chip being laid out by hand' they know that ultimately their solution, for one reason or another, has not worked exactly to plan and that a good full custom 28nm design can beat the crap out of it. Such solutions exist.


 How is every transistor on the new chip being laid out by hand a "strange announcement"? Hello, welcome to the definition of FULL CUSTOM DESIGN.

Care to explain how a full-custom 28nm design like the BM1385 used in the S7 can compete effectively after the halfing against a chip that has demonstrated over twice the efficiency and should still be PROFITABLE by then (it's looking more and more likely that the S7 is going to hit "unprofitable" shortly before the halfing unless you have VERY VERY cheap or FREE electric).

BitFury's real competiton for their new chip isn't going to be 28nm. It is going to be the upcomming A3, and at some point probably a 14/16nm full-custom chip from Bitmain, and possibly 1 or 2 others eventually going with full-custom at 14/16nm.


I get fed up saying this but people should read a lot more before shooting their mouths off. No one has laid a chip out by hand since the late 1970's - probably about the time your parents were born. This statement was just another piece of bullshit from Bitfury trying to make their chip sound 'special' in some way.

If you want some real, actual informed data about what full custom actually entails then I'm happy to recommend some very good books to you to help reduce your level of ignorance, you clearly don't really understand what full custom means or entails or what good engineers can do with it.

I'll bet that Bitmain make a lot of money on their S7's and could probably reduce it's price to sub $600 and still make a profit, so they have no real need in the near future to make a new chip (although I'm sure they will). They'll continue to make money on their 28nm cash cow for some time. They might even conjure up a containerised system of their own......

member
Activity: 233
Merit: 10
Punin's gotten in touch with me. There's no feasible way I'll get a million bucks but he's working on making small orders possible (if only in a roundabout way) which is definitely nice.

Good news. I would love to purchase some of your products with these chips.
sr. member
Activity: 441
Merit: 250
1. They are losing a lot of money or are under pressure from their investors to make more and need to
    make it up from the suckers, sorry, customers who are desperate to get their paws on the wonder chip
They are not.

Dogie, you're great with these short sentences which subtly suggest inside knowledge, but nless you have access to Bitfurys detailed management accounts then you have no idea what's actually going on, have you?. Yes, they have 16% of the network capacity which should make them about $6m a month net of power costs at BTC=$400, but they have a lot of mouths to feed - just look at the number of executive officers - 12 of them (!) plus another 5 board members and good knows how many hangers on not to mention the actual workers. Then they have to finance development and running costs and presumably they have raised $60M + by making some very challenging promises to their backers.

Why do you think they are selling their Golden Geese?  
sr. member
Activity: 324
Merit: 250
Philip: Wouldn't is be possible with immersion cooling with 3M novec 7000 in a 2 phase system?

legendary
Activity: 3892
Merit: 4331
Re container: probably a good deal if it costs 1-2mil and you have spare 2MW available.
I calculate that with halving coming and assuming just 2X difficulty increase until july 1 and 2X difficulty increase from July1 to Dec31 maximum bitcoin revenue would be somewhere between $4-5 mil in 12mo and ~$0.9-1.8 mil in electricity cost per year (first number for $0.05/kwh and second for $0.1/kwh).
However, who knows how much difficulty will increase. On the upside, price can increase and bail you out.
I wonder if anybody on the forum have access to 2mW?


I have access to 2MW.


and how will you handle  2MW of heat?

I see the top of that container and its fans giving off 4 million btus on the low end and 8 million btus on the high end

the container has about 40 by 10 foot area  so my heat disposal estimates tell me the fans won't be able to move 8 million  btus .  if anyone can tell me more info on this I would love to hear about the heat from those fans.

Maybe forum members should pitch in and form a co-op or LLC to buy one of these 2mW containers hosted at Dalkore if he is interested.
Perhaps 100 shares/slots with some trustworthy individual like Phil to handle payouts.
I can envision people able to invest $10K (maybe 15k or so) participating.
Of course, we would need to know the exact numbers and guarantees as it might be rather thin margins over the whole time.
1/100 of 16ph=160 Th. Right now you pay $34k for that much hashing power, so projecting forward the price for container should be realistically around 1 mil, and not significantly more.
There are too many moving parts, but it could be done, I guess.
legendary
Activity: 4256
Merit: 8551
'The right to privacy matters'
Re container: probably a good deal if it costs 1-2mil and you have spare 2MW available.
I calculate that with halving coming and assuming just 2X difficulty increase until july 1 and 2X difficulty increase from July1 to Dec31 maximum bitcoin revenue would be somewhere between $4-5 mil in 12mo and ~$0.9-1.8 mil in electricity cost per year (first number for $0.05/kwh and second for $0.1/kwh).
However, who knows how much difficulty will increase. On the upside, price can increase and bail you out.
I wonder if anybody on the forum have access to 2mW?


I have access to 2MW.


and how will you handle  2MW of heat?

I see the top of that container and its fans giving off 4 million btus on the low end and 8 million btus on the high end

the container has about 40 by 10 foot area  so my heat disposal estimates tell me the fans won't be able to move 8 million  btus .  if anyone can tell me more info on this I would love to hear about the heat from those fans.
legendary
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
If you figure all they have right now are engineering samples, could be a while before they have any to hand out to third parties. I think "a month or two" has been the going assumption for when they have actual production chips.
legendary
Activity: 3808
Merit: 1723
Punin's gotten in touch with me. There's no feasible way I'll get a million bucks but he's working on making small orders possible (if only in a roundabout way) which is definitely nice.

Any timeframe when you might receive these chips ?
sr. member
Activity: 324
Merit: 250
That would be perfect actually.
A rather silent 4TH would be awesome, and a 1TH pod is the dream! Wink
legendary
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
You and me both. A 500W 4TH kit would be pretty smexy. Also a 1TH pod and 100GH stick.
sr. member
Activity: 324
Merit: 250
Punin's gotten in touch with me. There's no feasible way I'll get a million bucks but he's working on making small orders possible (if only in a roundabout way) which is definitely nice.

Good to hear you are involved. I'd really like to see you making a miner with these chips!
legendary
Activity: 3374
Merit: 1859
Curmudgeonly hardware guy
Punin's gotten in touch with me. There's no feasible way I'll get a million bucks but he's working on making small orders possible (if only in a roundabout way) which is definitely nice.
hero member
Activity: 588
Merit: 500

 I also suspect BitFury's claim about "widely available" is STILL talking about that lightbulb thing of theirs.

I think it could go beyond this as I am expecting to see a number of 3rd party Miners based on the BF8162C16 chip. They seem to be actively recruiting for "Integrators" http://bitfury.com/products#application-form so am hopeful that there will be a number of options? Let's hope there will be some affordable Miners and that one of them will be Sidehack?

As to the lightbulb I would still like to see it, as hopefully I will be able to afford one and it would be nice to have something low cost to take apart & play with.  Smiley


Rich
legendary
Activity: 1498
Merit: 1030

Like everything that Bitfury announces you have to read very carefully, I refer to: "We understand it will be nearly impossible for any older technology to compete .......". The 'nearly' is the giveaway, despite their strange announcement about 'every transistor on the new chip being laid out by hand' they know that ultimately their solution, for one reason or another, has not worked exactly to plan and that a good full custom 28nm design can beat the crap out of it. Such solutions exist.


 How is every transistor on the new chip being laid out by hand a "strange announcement"? Hello, welcome to the definition of FULL CUSTOM DESIGN.

 Care to explain how a full-custom 28nm design like the BM1385 used in the S7 can compete effectively after the halfing against a chip that has demonstrated over twice the efficiency and should still be PROFITABLE by then (it's looking more and more likely that the S7 is going to hit "unprofitable" shortly before the halfing unless you have VERY VERY cheap or FREE electric).

 BitFury's real competiton for their new chip isn't going to be 28nm. It is going to be the upcomming A3, and at some point probably a 14/16nm full-custom chip from Bitmain, and possibly 1 or 2 others eventually going with full-custom at 14/16nm.


I discount KnC's Solar, if they had THAT in full production they'd be showing a lot higher hashrate climb than they've demonstrated since their last annoucement about the thing.



 I also suspect BitFury's claim about "widely available" is STILL talking about that lightbulb thing of theirs.
legendary
Activity: 1330
Merit: 1026
Mining since 2010 & Hosting since 2012
Re container: probably a good deal if it costs 1-2mil and you have spare 2MW available.
I calculate that with halving coming and assuming just 2X difficulty increase until july 1 and 2X difficulty increase from July1 to Dec31 maximum bitcoin revenue would be somewhere between $4-5 mil in 12mo and ~$0.9-1.8 mil in electricity cost per year (first number for $0.05/kwh and second for $0.1/kwh).
However, who knows how much difficulty will increase. On the upside, price can increase and bail you out.
I wonder if anybody on the forum have access to 2mW?


I have access to 2MW.
legendary
Activity: 4592
Merit: 1851
Linux since 1997 RedHat 4
1. They are losing a lot of money or are under pressure from their investors to make more and need to
    make it up from the suckers, sorry, customers who are desperate to get their paws on the wonder chip
They are not.

2. They are worried that Bitmain have got a very cost effective solution which will ultimately come to
    dominate the network before they can get any real advantage from 16nm. I'm guessing that as well as
    mining Bitmain make very healthy margins from their S7's
If they're not, they should be. Bitfury has a pretty terrible history of dealing with anyone who doesn't have a representative lawyer where as Bitmain specifically caters from the one personal machine up to the corporate megafarm.

If my cater you mean go on vacation with no one at the office for two weeks to do repairs, RMAs, ship, etc... lol
... and they cater for it so well with dodgy software that's not been fixed for years ...

Also, it seems that the S7 includes the security risk extranonce subscribe hack also ...
member
Activity: 98
Merit: 10
1. They are losing a lot of money or are under pressure from their investors to make more and need to
    make it up from the suckers, sorry, customers who are desperate to get their paws on the wonder chip
They are not.

2. They are worried that Bitmain have got a very cost effective solution which will ultimately come to
    dominate the network before they can get any real advantage from 16nm. I'm guessing that as well as
    mining Bitmain make very healthy margins from their S7's
If they're not, they should be. Bitfury has a pretty terrible history of dealing with anyone who doesn't have a representative lawyer where as Bitmain specifically caters from the one personal machine up to the corporate megafarm.

If by cater you mean go on vacation with no one at the office for two weeks to do repairs, RMAs, ship, etc... lol
Pages:
Jump to: