state of mining after asic arrival?

[arklan]

i keep getting stuck when i try to think about where the "mining industry" as it were is all going to wind up after ASIC hardware actually arrives (bfl or otherwise). i just can't wrap my brain around it. so, here's a thread. 

even with the very low cost of running asic's, the number of people willing to do so at a loss is, i think we can safely agree, minimal compared to those interested in making money from mining. it seems to me the whole idea of block rewards and transactions fees going to miners is specifically geared toward a profitable venture to ensure people are there mining to provide network security and to actually, you know, confirm the transactions.

but if it's not profitable to mine, people will turn off their rigs- asic or not. in doing so, transaction confirm times will climb until the difficulty drops at the next retarget. at that time, the difficulty will drop making it profitable to mine again, and thus they'll switch them back on confirm times will decrease until the next retarget and so on, ad nauseam. how would this sort of cyclic variance effect network security and general faith in bitcoin? one week a transaction confirms in the expected ten minute, then next in five, the next in twenty or thirty minutes as the total network hashrate fluctuates wildly.

there's plenty more headache inducing confusion on my end from this ASIC stuff... but i can't seem to actually form coherent thoughts from it, so let's just go with this for now.

so, have i wildly misinterpreted things? will everyone just leave their asics on, regardless of profit or loss? was mining never meant to be profitable? will i survive the rest of the year without my brain exploding and leaking out my ears?

[Dargo]

We haven't seen the radical fluctuations in the network you are imagining pre-ASIC, and I don't see any reason to think we will see them post ASIC. Basically people vary widely in terms of their judgments about whether it is "worth it" to continue mining at the present moment or not, so people aren't going to be turning their rigs on and off en masse as you are imagining. 

[Meni Rosenfeld]

As Dargo says it's going to converge to an equilibrium, where the difficulty and total hashrate are just right for each other. Very specific circumstances are required for the kind of oscillations you speak about.

Here's a model: Let's say the network total hashrate (affected by how profitable it is to mine with existing equipment, and measured in shares per second) is H = a - b*D, where D is the difficulty, assumed to be close to the equilibrium difficulty. The retargeting algorithm basically measures H and computes D based on it, so D_{n+1} = 600*H_n = 600*(a - b*D_n). This converges to equilibrium as long as b<1/600. For b to be higher than this, the cost of running equipment needs to be significant compared to the cost of buying it (which shouldn't be the case for ASIC for some time), and furthermore, to be homogeneous between all miners.

[arklan]

well that allays that worry. thanks guys.

[johnyj]

I remember the I0COIN public lunch, in the first 5 minutes I could mine several blocks and the second day I could only mine several coins

Same thing will happen to ASICs, during the first several batches the difficulty will rise quickly but eventually it will stablize and only those have the long term patience will still carry on

There is a potential problem: Today many miners are using gaming GPUs to do mining, they do this easily without any other investment. But if ASIC comes out and GPUs are not capable of generate meaningful coin anymore, they might just quit mining and forget about BTC, this will in turn dramatically reduce the popularity and the support of BTC, unless ASIC devices can be purchased everywhere like GPU today. FPGA's performance are on par with GPUs, so they never made any threat to GPUs, but ASICs are different

[arklan]

There is a potential problem: Today many miners are using gaming GPUs to do mining, they do this easily without any other investment. But if ASIC comes out and GPUs are not capable of generate meaningful coin anymore, they might just quit mining and forget about BTC, this will in turn dramatically reduce the popularity and the support of BTC, unless ASIC devices can be purchased everywhere like GPU today. FPGA's performance are on par with GPUs, so they never made any threat to GPUs, but ASICs are different

i think that's exactly what BFL is attempting to solve with the jalepeno product - $150 is cheap even for a video card. i could easily sell my ati 6870 i use for mining and get a jalepeno at almost no cost, giving me more then a 10x increase in hashrate (assuming they actual deliver on price/preformance) if the difficult increased by a factor of ten, i'd more or less be right where i am now, profit wise. maybe even better since the electricity costs would be lower. thus any interested GPU miners could sell their hardware they aren't otherwise using, get a jalepeno and carry on like nothing happened.

course if it's a 50x difficulty increase it would be a different story, and myself and i believe others don't want to buy ado business with BFL with their way of doing business... man we need competition in the ASIC realm...

[bpd]

ASIC should improve the situation, since the variable (electricity) cost of running a rig drops dramatically. Most of the cost of mining would now be fixed (capital) cost. BTC price and difficulty should influence mainly the volume of new capital investment, not whether existing rigs are running.

[ElectricMucus]

Quite simple: The performance won't be anything like the initial promises, like last time.

If we keep the previous ratio: The BFL single was said to have 1000 mh/s and 20 W instead it turned out to be 80 W and 800 mh/s
Granted 2.8 gh/s (80%) and 20 W (400%) is still nothing to sniff at but it won't be this "little" exaggeration this time.

I've gone over some length on another thread to explain to people that the BFL ASIC isn't a full custom device and more in the ballpark of Altera Hardcopy.
These are the same devices as their FPGAs but with a metal layer instead of the routing fabric. The resulting benefits are higher clock rates, lower power consumption and a cheaper price. But they aren't as great as you might think and certainly not 100+% in clock rate and 1.5% (!) in power consuption.

Something realistic for a "cup warmer" would be 1.6gh along with 40W power. (If you like boiling coffee you'll be fine )

[arklan]

oh yea. i don't expect BFL to really meet their promised preformance, at least not even close to october...

i'm just trying to figure out how drastic the change will be when full custom ASIC's do arrive. right now i'm squeezing about 330 mhash out of my setup (i know, it's sad...) but over time i could buy, for example, a modminer backplane and 1 chip for $350 usd, then slowly expand and such. if difficulty were relatively stable (as it has been, more or less) then eventually i could live off the mining profits, even at the current exchange rate.

but that's all pointless if ASIC's or other dramatic difficulty increasing hardware is arriving and wiping out the value of the hardware i had.

i really need to persue a bitcoin business instead... shame i can't think of anything.

[johnyj]

but that's all pointless if ASIC's or other dramatic difficulty increasing hardware is arriving and wiping out the value of the hardware i had.

GPU is almost riskless investment since you can still use those high end GPUs to play games for couple of years

GPUs have already wiped out those CPU mining possibilities, due to it's almost 40X efficiency increase, just hope after many people have merged to ASIC mining, there are no such dramatic increase in efficiency again

[swissmate]

GPU miners will just start mining LiteCoins, It is already more profitable now and when ASIC comes out, they wont have an option.

So true

[Dargo]

Something realistic for a "cup warmer" would be 1.6gh along with 40W power. (If you like boiling coffee you'll be fine )

I don't think they will be this far off the estimated specs. But even if they are, for $150 this still beats the pants off anything else available.

[nedbert9]

Open question.


Would you buy a Jalepeno if it made $0.67 per day?

[DeathAndTaxes]

Would you buy a Jalepeno if it made $0.67 per day?

Yes to help decentralize and protect the network.   I would imagine a lot of people would.  I won't however pay for one and wait x months till BFL gets around to actually delivering so I will be waiting probably until 2013.

[arklan]

Would you buy a Jalepeno if it made $0.67 per day?

Yes to help decentralize and protect the network.   I would imagine a lot of people would.  I won't however pay for one and wait x months till BFL gets around to actually delivering so I will be waiting probably until 2013.

exactly this. .67 USD a day is about 20 a month, which, being an ASIC, should beat the electric cost just fine. and as long as that's the case, so be it. but like D&T says, i'm not a fan of the whole preorder-for-full-price thing. if i buy anything from BFL, there will have to be a MUCH more reliable and concsistent shipping timeframe.

[johnyj]

ASIC should improve the situation, since the variable (electricity) cost of running a rig drops dramatically. Most of the cost of mining would now be fixed (capital) cost. BTC price and difficulty should influence mainly the volume of new capital investment, not whether existing rigs are running.

I'm also looking into this aspect, I think that sooner or later electricity will again become the deciding factor for profitability

Today a miner use FPGA to mine 1 BTC/day with 10% paid for electricity, and next year he will use ASIC miner and network difficulty will rise accordingly, he still mine 1 BTC/day with almost the same electricity cost. If the electricity cost so little, he could easily expand his operation 10X to get 10 BTC per day, and everyone else will also do the same, together they will push the difficulty to 10X level, at that time, the electricity for that miner will cost 1BTC/day, but return stays at 1BTC/day unchanged

ROI is not important, but profitability is important. When there are plenty of capital (I heard that QE3 is coming) can flow into the market to chase those 3600 BTC,  there will be mass produced ASIC miners everywhere, and their price will drop quickly. But when the electricity cost rise to a level that most of the miner can not operate profitably, the total network hashing power will be stablized

[arklan]

that's when the serious miner's invest in solar, wind and other such forms of "free" electricity.

[GernMiester]

D and T,
I say no way.
Most people simply don't care about the network and want money that's it.  I am sure a few said this is great I want to support this idea. The majority of people say I want money, I want it fast, and I don't care about. If that isn't the case why did the Nigerian prince scam ever take off. The rest say they care about the network but its a lie and all they really want the cash.

Seriously, do you think BTC would even matter if it could not be converted to FIAT somehow. BTC is not new and it barely accepted anywhere.

[check_status]

If ASIC use explodes, then difficulty will be effected. An increasing difficulty would reduce a GPU miners ability to remain profitable.
BFL alone can't supply enough ASIC's to create such a catastrophic change in difficulty which would harm the GPU miners, initially. Within a year of the ASIC's release though, it could begin to have an impact upon difficulty where GPU miners realize their profit declining.
How many additional terahashes would it take to increase difficulty?
What level of difficulty would make GPU mining an unprofitable endeavor?
How many ASIC units can BFL deliver per month, per year?
For every 1000 Jalapenos sold that's 3.5 terahashes, for every 10,000 that's 35 terahashes.

For $500 I can have 14 Gigahash, I would buy 4 right now even if I won't get them for 4-6 months.