20ph increase in btc mining network

[ihasher]

Where did it come from? Sp30s?

No. It's BitFury.

Do you KNOW it's BitFury or speculating?

I know.

Something doesn't add up.

You claim to know about their chips. You claim to know about their datacenters. You claim that they brought up 20MW in Iceland. However, almost 40PH/s just hit the network.

So, the question that all of your customers want answered: Are they using your chips or not? Otherwise how do you explain 40PH/s?

[philipma1957]

1)you would be competing against yourself.
 

You are always competing with yourself, whether you add 300 PH/s or 1 GH/s to the network.  That is something that won't change if you add the hash power at a slower rate.  There is no advantage to adding it slower from what I can see.

well at the scale of 160ph + 300ph  = 460ph

adding 300 ph at 1 watt per hash is quite a task.   1th = 1k-watt  1 ph = 1m-watt

 300 ph = 300 megawatt

 building the 300 ph in mining gear is not that costly.

having the cooling and the power is harder.

at a certain point large scale mining gear production can be very low cost,

 but to run the mining gear you need access to cheap power.  

 the Niagara Falls power plant usa + can  could run  5,000 m watts

that means a plant is build able  but how many areas have the right setup to make a 300 m watt plant

the idea that I can make gear cheaply.  then add it on as I access power cheaply is not far fetched as some think.

here is a link to Niagara falls power cap.

http://www.niagarafrontier.com/power.html  2700 mw USA + 2300 mw CAN = 5000mw

 enough to run 30x the current network if we burn 1 watt per gh

or 15x the current network if we burn 2 watts per gh

 If bitfury and bitmaintech get access to 2 cent a kwatt for power in large amounts they will build big places.

Not complicated it will happen. 

[SMB-2525]

This article thinks the Republic of Georgia has about 2.3 GW of exploitable hydropower. Not inexpensive to bring on line:

http://www.hydroworld.com/articles/print/volume-21/issue-01/articles/russia---central-asia/georgian-hydropower-turns-river.html

BPA tariff to a public utility district (mostly in Washington state) is less than 0.04/KWH before transmission and other markups. A lot of those dams were build in 30s-60s and so capital cost is largely amortized already.

[SMB-2525]

 enough to run 30x the current network if we burn 1 watt per gh

or 15x the current network if we burn 2 watts per gh

 If bitfury and bitmaintech get access to 2 cent a kwatt for power in large amounts they will build big places.

Not complicated it will happen. 

At 0.02/KWH, 16% bumps every 12 days one Watt/GHS units go negative on power in 5/18/2015. .85 Watt/GHS units go negative 5/30/2015. .5 Watts/GHS units go negative 7/17/2015.

Something is going to change.

[SMB-2525]

Another future perspective @ .85 WATT/GHS form goes online at the 11/7/2014 bump date. Same assumptions ($600, 16%, 12 days).
Difficulty is 64681965063

The plant goes negative on power alone by 5/18/2015 after producing $290 worth of BTC per tera-hash.

Consider the ASIC team planning the .5W/GHS chips. Assume those farms can go online in time for 3/31/2015 bump. Difficulty is 383953893734.

That plant will go negative on power in 108 days after producing $24.50 per terahash of bitcoin (7/5/2015)

Something is going to change.

[philipma1957]

Another future perspective @ .85 WATT/GHS form goes online at the 11/7/2014 bump date. Same assumptions ($600, 16%, 12 days).
Difficulty is 64681965063

The plant goes negative on power alone by 5/18/2015 after producing $290 worth of BTC per tera-hash.

Consider the ASIC team planning the .5W/GHS chips. Assume those farms can go online in time for 3/31/2015 bump. Difficulty is 383953893734.

That plant will go negative on power in 108 days after producing $24.50 per terahash of bitcoin (7/5/2015)

Something is going to change.

yes and this brings about my point of building the gear  already and releasing it into the network slowly.

  If you build 300ph for  cheap.  say 30 million and the power plant will be 90 million to build. 

 you can add the hash slower not in a big dump.  Should be interesting to see if  china and russia end up with 2 centers each one with 200-300ph.

Bitmaintech was selling s-1 shipped to the usa as low as 172 usd.  lets say it cost them 100 bucks of which 25 was shipping. so 1 piece was 75 usd.

So if a s-3 = 75 usd to build  21 = 10th  210 = 100th

2100 = 1 ph  so they can build 1 ph for 2100  x 75 =  or 157,500 usd    that is 1,575,000 usd for 10ph  and 15,750,000 usd  for 100 ph


all ball park but not crazy wrong. so making 200 ph may be  doable for 30 million.    remember they could under clock it and use .5 watts  so  making  gear and storing it is

not as crazy as it seems.   the gear alone  is just one part of the data center.    if you build an s-3 center for 21000 units it is 10 ph

 so a 30 ph center is 63,000 units 

[dropt]

Something doesn't add up.

You claim to know about their chips. You claim to know about their datacenters. You claim that they brought up 20MW in Iceland. However, almost 40PH/s just hit the network.

So, the question that all of your customers want answered: Are they using your chips or not? Otherwise how do you explain 40PH/s?

They're not using SP-Tech chips.  The bitfury group has their own desing which also happens to be the best 55nm chip to have ever hit the market.  In fact it competes on the same level as other manufacturers 28nm.

So to put it succintly.  The Bitfury group, who opened up the 20MW/40PH farm is using their own hardware with their own (and VC) money.

[Hfleer]

It looks like longterm only the chip producers themselves will be able to mine profitably. Hopefully none of them gets an extreme advantage in technology, otherwise centralisation is inevitable.

[philipma1957]

Something doesn't add up.

You claim to know about their chips. You claim to know about their datacenters. You claim that they brought up 20MW in Iceland. However, almost 40PH/s just hit the network.

So, the question that all of your customers want answered: Are they using your chips or not? Otherwise how do you explain 40PH/s?

They're not using SP-Tech chips.  The bitfury group has their own desing which also happens to be the best 55nm chip to have ever hit the market.  In fact it competes on the same level as other manufacturers 28nm.

So to put it succintly.  The Bitfury group, who opened up the 20MW/40PH farm is using their own hardware with their own (and VC) money.

so you think they are getting .5 watts a hash?

20 mw for 40 ph comes to .5 watts   a gh .  I am thinking that chip can do under 1 watt say .9 not .5 watts

the 40 ph may be a bit less say 20-25ph bitfury and  some sp30's alog with some s-3's and a bit of good luck

still may top us over 20% .  I wonder if we get 3 under 10% after this big jump

[SMB-2525]

20 mw for 40 ph comes to .5 watts   a gh .  I am thinking that chip can do under 1 watt say .9 not .5 watts

Do we know of any .5 WATT/GHS setup?

[philipma1957]

20 mw for 40 ph comes to .5 watts   a gh .  I am thinking that chip can do under 1 watt say .9 not .5 watts

Do we know of any .5 WATT/GHS setup?

no confirmed tested gear for the little guy does .5 watts

the sp30 is .6 watts not sure if dogie has done his review on it.

[jonnybravo0311]

No way is BitFury getting 0.5W per GH/s on that 55nm chip.  I'd say the increase is a combination of hardware from BitFury, Spondoolies-Tech and Bitmain and good old fashioned variance.

[davejh]

No way is BitFury getting 0.5W per GH/s on that 55nm chip.  I'd say the increase is a combination of hardware from BitFury, Spondoolies-Tech and Bitmain and good old fashioned variance.

We're certainly well within the normal ranges for variance. Looking at the log plots:

https://blockchain.info/charts/hash-rate?showDataPoints=false&show_header=true&daysAverageString=1&timespan=&scale=1&address=

The previous difficulty period looks like it was a pretty unlucky period (could be statistical variance, could be hardware was offline and is now back online) and we're now seeing things returning to the sort of trajectory that they probably should have been on all along.

[jonnybravo0311]

No way is BitFury getting 0.5W per GH/s on that 55nm chip.  I'd say the increase is a combination of hardware from BitFury, Spondoolies-Tech and Bitmain and good old fashioned variance.

We're certainly well within the normal ranges for variance. Looking at the log plots:

https://blockchain.info/charts/hash-rate?showDataPoints=false&show_header=true&daysAverageString=1&timespan=&scale=1&address=

The previous difficulty period looks like it was a pretty unlucky period (could be statistical variance, could be hardware was offline and is now back online) and we're now seeing things returning to the sort of trajectory that they probably should have been on all along.

I completely agree that we're in the normal range for variance.  However, we also know that at least *some* amount of hardware has come online in the past week - Spondoolies-Tech SP30s are mining (last I heard was there were about 1000 August units - but not entirely sure how accurate that number is, or how many of those are actually currently online).  Supposedly BitFury has fired up the new data center and dropped some hardware in.  Don't forget Bitmain has sent quite a few S3s out.  Anyway, the point I was making is that BitFury isn't getting 0.5W per GH/s, and isn't solely responsible for the increased hash rate.

[xstr8guy]

There is a lot of conjecture here on what Bitfury is doing with their hardware but the truth is... we may never know since they are not releasing their new chip to the public.

The original 55nm chips did under 1w/GH. But they did a respin/redesign on the 55nm which had some dead engines in the original chip. And perhaps they were able to improve it by a considerable margin. Now just imagine if they could shrink the die size to 28nm or lower. And they certainly have the money to be able to afford a 14nm chip. They'd leave every competitor in the dust .

[SMB-2525]

There is a lot of conjecture here on what Bitfury is doing with their hardware but the truth is... we may never know since they are not releasing their new chip to the public.

The original 55nm chips did under 1w/GH. But they did a respin/redesign on the 55nm which had some dead engines in the original chip. And perhaps they were able to improve it by a considerable margin. Now just imagine if they could shrink the die size to 28nm or lower. And they certainly have the money to be able to afford a 14nm chip. They'd leave every competitor in the dust .

The problem they face is when it rolls out and how much it costs to put into production. I ran the numbers on a 0.1 Watt/GHS rig rolling out on 1/1/2015.
@ 0.02KWH, $600/BTC and the current 2014 average of 16% and 12 days per bump, it cash flows on power only for 312 days. A 1 THS rig will produce 1/3 BTC before going negative on power only. Starting difficulty was 1.17116E+11 which is today increasing at 16% every 12 days.

If they wait until 3/1/2015 to go into production with the 1Watt/GHS rig, that same 1 terahash rig will produce 1/6th of a BTC.

I don't see the business case for a major new investment in mining technology. It is a prisoner's dilemma http://en.wikipedia.org/wiki/Prisoner's_dilemma type of problem. If everyone stops investment, the value of the existing investment is maximized. If one entity keeps investing, the value of everyone's investment decreases.

In the years to come, I think the economic behavior of the big BTC players over the next 12-18 months will be studied in great detail. It is a mathematically bounded problem where the outcomes only depend on the behavior of the other big players.

[cloverme]

What gets turned off will likely get turned back on... heh... I don't think we've seen the last of our 20ph "friend".    I doubt we will ever know what it was or who it belonged to.

[Bitsaurus]

The hashrate seems to be accelerating even more so today as the block solves in the last 24 hours is exceeding 200 now.  Could be luck, could be lots of new eq.  I haven't bothered to look at all the pools to see if it's luck.

[SMB-2525]

The hashrate seems to be accelerating even more so today as the block solves in the last 24 hours is exceeding 200 now.  Could be luck, could be lots of new eq.  I haven't bothered to look at all the pools to see if it's luck.

Now done to 173 and block time up over 8.5 minutes. Jumpy.

[taipo]

Hopefully none of them gets an extreme advantage in technology, otherwise centralisation is inevitable.

Centralisation is already happening as we speak.

At the current rate, by the end of the year the majority of the hashrate will come from a dozen or more large corporations running multiple data centres around the globe. They will probably go at each other for a period until a few drop off, and the equilibrium will probably be spread across  about 5 or 6 major corporations throughout 2015 - 2016.