Bitcoin difficulty is slowing down?

[jenga]

For some reason, according to this chart, the bitcoin network hashrate is having major fluctuations and it appears as it is losing its exponential growth:
http://bitcoin.sipa.be/speed-lin.png
Why is that? Some big pool is having issues lately?

[DeathAndTaxes]

For some reason, according to this chart, the bitcoin network hashrate is having major fluctuations and it appears as it is losing its exponential growth:
http://bitcoin.sipa.be/speed-lin.png
Why is that? Some big pool is having issues lately?

KNC stopped shipping?  Don't worry they will be shipping 2.5 PH/s+ pretty soon again.


Also you do understand that exponential growth forever is impossible right?  Within a couple years Bitcoin mining would use more energy than used by the human race for all other purposes combined.  Law of large numbers takes over.   When network is 1 PH/s for the network to double in a month requires 1 PH/s more hash power today it would require >5 PH/s so even if for example ASIC companies shipped out constant volume the growth as a % would decline.  For the growth % to remain constant they would need to manufacture and ship an exponential volume.  Thousands of units per month this year,  millions of units per month next year, billions of units per month the year after that, trillions of units per month the year after that, quadrillions of units per month the year after that ...

[jenga]

KNC stopped shipping?  Don't worry they will be shipping 2.5 PH/s+ pretty soon again.

Acknowledged.

Also you do understand that exponential growth forever is impossible right?  Within a couple years Bitcoin mining would use more energy than used by the human race for all other purposes combined.  Law of large numbers takes over.   When network is 1 PH/s for the network to double in a month requires 1 PH/s more hash power today it would require >5 PH/s so even if for example ASIC companies shipped out constant volume the growth as a % would decline.  For the growth % to remain constant they would need to manufacture and ship an exponential volume.  Thousands of units per month this year,  millions of units per month next year, billions of units per month the year after that, trillions of units per month the year after that, quadrillions of units per month the year after that ...

That's true of course. But one shall also consider the efficiency of new ASIC chips making the exponential growth last longer. If even asic companies sell at a constant rate, but their chips become faster and more efficient the effect is the same with lower energy consumption. As older mining rigs get replaced by new less power hungry miners the total network energy consumption may even decrease (very unlikely but hey, we are in the "speculation" subsection ).
My point is that the hashrate will continue going up exponentially for an unexpectedly long amount of time.

[DeathAndTaxes]

On the efficiency aspect it is nothing compared to the prior difficulty increases.

A GPU is something like 300 J/GH.  The "worst" ASIC is ~8 J/GH so you are talking almost a 40,000% improvement.   Various different smaller process node designs dropped from the 8 J/GH to 5 J/GH to ~0.8 J/GH (Bitfury - reference clock)).

On paper we have the rest of the 28nm players showing ~0.8 J/GH and KNC estimating ~0.7 J/GH.   Seeing a trend.  Overnight we went from 300 J/GH to 8 J/GH and then over the next six months saw that fall by a factor of 10 to ~0.8 J/GH.   That "might" go down to a staggering 0.7 J/GH by next Spring.

My point is that the hashrate will continue going up exponentially for an unexpectedly long amount of time.

Who is going to buy all that hashrate increases. $3 per GH works out to $3M per PH.   So to double the network again in a month would "only" require $15M in sales.  Ok no problem there.  To double it again would require another $30M in sales the following month, $60M the next month, $120M the monh after that, a quarter billion the month after that ...

Will difficulty keep going up?  Oh HELL YEAH!  Is it possible from either a cashflow or energy standpoint to double continually every month for years?  No.

[CobaltBlueD]

On the efficiency aspect it is nothing compared to the prior difficulty increases.

A GPU is something like 300 J/GH.  The "worst" ASIC is ~8 J/GH so you are talking almost a 40,000% improvement.   Various different smaller process node designs dropped from the 8 J/GH to 5 J/GH to ~0.8 J/GH (Bitfury - reference clock)).

On paper we have the rest of the 28nm players showing ~0.8 J/GH and KNC estimating ~0.7 J/GH.   Seeing a trend.  Overnight we went from 300 J/GH to 8 J/GH and then over the next six months saw that fall by a factor of 10 to ~0.8 J/GH.   That "might" go down to a staggering 0.7 J/GH by next Spring.

My point is that the hashrate will continue going up exponentially for an unexpectedly long amount of time.

Who is going to buy all that hashrate increases. $3 per GH works out to $3M per PH.   So to double the network again in a month would "only" require $15M in sales.  Ok no problem there.  To double it again would require another $30M in sales the following month, $60M the next month, $120M the monh after that, a quarter billion the month after that ...

Will difficulty keep going up?  Oh HELL YEAH!  Is it possible from either a cashflow or energy standpoint to double continually every month for years?  No.

No argument the rate of increase is unsustainable.  Of course, $/GHs and J/GH will decrease. I haven't run the numbers but going back to CPU/GPU days it's pretty obvious that constant investment will still result in significant increase in difficulty. Mining ROI is painful.

[Syke]

This is the calm before the storm. Enjoy it while you can.

[Sharky444]

This is the calm before the storm. Enjoy it while you can.

+1

Next year large corporation will go into Bitcoin mining. Home users will loose the possibility to mine.

[TheJuice]

We will see ~40% increase per month until May of the next year. Beyond that I see 2 options - complete plateau or BTC goes up enough in price/we go mainstream that a 'real' company (intel, etc) starts making mining equipment for the masses.

Things are finally looking up for miners.

[redmetal]

We will see ~40% increase per month until May of the next year. Beyond that I see 2 options - complete plateau or BTC goes up enough in price/we go mainstream that a 'real' company (intel, etc) starts making mining equipment for the masses.

Things are finally looking up for miners.

I tend to agree with this, 40% for a little while seems fesable, Cointerra, Hashfast, VMC, and KNC for their second round will continue this boost,
Look at the lastest GPU's 11nm is where we are at, I don't see any company speeding the 10's of millions required to tape this out, not now anyway, maybe when bitcoin is mainstream, but why would a company like intel want to share in this small time show? ASIC companies are only selling millions worth or hardware, intel would need to be purswaded by 100's of millions before pumping an investment that large into something this small.

[DeathAndTaxes]

Look at the lastest GPU's 11nm is where we are at

Are you from 5 to 7 years in the future or are these GPU from imaginary land?

Top of the line GPUs are 28nm today.

[Bitcoinorama]

Oh no, so you mean all the FUD the Genesis block spreads is just to redirect traffic to their website as part of some marketing scam??

Who'd have thought?

[redmetal]

Look at the lastest GPU's 11nm is where we are at

Are you from 5 to 7 years in the future or are these GPU from imaginary land?

Top of the line GPUs are 28nm today.

Apologies DRAM is at 11nm gpus are almost at 20nm

[AMuppInTime]

ASICminer plans a chip with <.2J/GH for January 20 Tapeout... Specs expected before tapeout. They are one of the big HW manufacturers and decided to skip their gen2 to focus of gen3... And have the experience too.
(discl - holder). friedcat (CEO) assumes the network difficulty will be 1000PH in June...

[sgbett]

For some reason, according to this chart, the bitcoin network hashrate is having major fluctuations and it appears as it is losing its exponential growth:
http://bitcoin.sipa.be/speed-lin.png
Why is that? Some big pool is having issues lately?

KNC stopped shipping?  Don't worry they will be shipping 2.5 PH/s+ pretty soon again.

BFL also cleared their backlog...

I'm enjoying the breather, I might get 3 BTC from my single instead of 2

[Motya]

For some reason, according to this chart, the bitcoin network hashrate is having major fluctuations and it appears as it is losing its exponential growth:
http://bitcoin.sipa.be/speed-lin.png
Why is that? Some big pool is having issues lately?

At the very end of that graph, aren't the estimate lines speeding up with a vengeance?

[DrG]

KNC batch 2 delivered on time

[Syke]

I'm enjoying the breather, I might get 3 BTC from my single instead of 2

A 30% difficulty adjustment increase is a breather?

[jenga]

For some reason, according to this chart, the bitcoin network hashrate is having major fluctuations and it appears as it is losing its exponential growth:
http://bitcoin.sipa.be/speed-lin.png
Why is that? Some big pool is having issues lately?

At the very end of that graph, aren't the estimate lines speeding up with a vengeance?

Not when I started the thread...

[sgbett]

I'm enjoying the breather, I might get 3 BTC from my single instead of 2

A 30% difficulty adjustment increase is a breather?

Me and my big mouth eh