Why You Haven't Seen Miners Leave in Hordes.....

[CanaryInTheMine]

Don't forget the botnets!  Since there is now malware out that can utilize a person's GPU to mine coins, I wouldn't be surprised if as much as 1/3 of all the TH/s is acquired through botnet activity.

1/3rd??? how do you figure that number?  that would be PR disaster for bitcoin...

[1714788360]

1714788360

[1714788360]

1714788360

[1714788360]

1714788360

[SgtSpike]

Don't forget the botnets!  Since there is now malware out that can utilize a person's GPU to mine coins, I wouldn't be surprised if as much as 1/3 of all the TH/s is acquired through botnet activity.

1/3rd??? how do you figure that number?  that would be PR disaster for bitcoin...

Just a guess.  But I wouldn't be surprised.

Antivirus agencies have already reported on systems being infected with GPU-mining trojans.  Since such viruses are out in the wild, who knows how many systems have been infected?  I think the last estimate was there was something like 50,000 GPU's making up the total amount of hashing power.  And there's how many computers in the world?  Several billion?  Granted, the vast vast majority of those do not have a graphics card capable of producing more than a few MH/s, but still, it wouldn't take much of a percentage of those billions of computers having a trojan to make up a decent amount of the total hashing power in the Bitcoin network.

Think about it...

[eldentyrell]

Just a guess.  But I wouldn't be surprised.

I would.  The vast majority of botnet machines are desktops in some office, for the simple reason that those are the vast majority of PCs.  These sorts of machines rarely have GPUs with any significant amount of hashpower.

[SgtSpike]

Just a guess.  But I wouldn't be surprised.

I would.  The vast majority of botnet machines are desktops in some office, for the simple reason that those are the vast majority of PCs.  These sorts of machines rarely have GPUs with any significant amount of hashpower.

Ok, what's the typical botnet size?  And we'll go from there...

[P4man]

There are more laptops than there are desktops. There are FAR more corporate style desktops than there are gaming rigs. I would be surprised if 1/10th of the PCs out there would be remotely suitable for mining.
Scratch that, more like 1/100, as it seems even among gamers,  only 5% has DX11 capable hardware:
http://store.steampowered.com/hwsurvey

[eldentyrell]

There are more laptops than there are desktops.

Another good point.  Can't believe I forgot about that.

Heat/power-sensitive laptops are great for sending "V14GR4 SP4MZ T0 3NL4RGE J00R M4NH00D!!!!" but pretty useless for mining.

[Vitalik Buterin]

Mining peaked around August 1, 2011.
The mining graph has roughly the shape of the Bitcoin/USD price graph, but trails it by about two months. One interpretation of this is that miners hang on an average of two months after they start losing money, then quit.

A proper economic analysis of this needs to take into account both the initial cost, buying the video cards, and the running cost (electricity). Back in June, miners thought that the price was going to keep on going up, so it was seen as worth it for a very large number of new miners to come in, and they did, paying the initial cost. The price went far down, and it is now no longer worth it for new miners to enter (unless they're not paying for their electricity), but even in July it was worth it for existing miners to keep mining.

In August, however, the price kept on falling, and fell below the threshold so that the least efficient of even the existing miners were forced to shut down, which is why it's only then that the number of miners started to go down. It's not a 2 month lag, it's http://en.wikipedia.org/wiki/Hysteresis - like a thermostat only increasing or decreasing the power to the heating when the temperature falls substantially below the target. The price threshold for miners to stay mining is about 1/3 of the threshold for there to be new miners, so there is a 3x range within which the price can change without changing the number of miners, but if the price falls outside the range it drags the number of miners up and down with it (technically, it's the expectation of future high prices that brings new miners in, and that's what happened in June, but it's a good simplification in the long term).

[AngelusWebDesign]

Regarding botnets GPU mining on the sly --

As someone pointed out, that's very difficult as many people would wonder why their office is suddenly 10 degrees hotter than it was last week, with the same outside temperature and/or household climate control settings.

"Last week, my A/C was set at 78 and my office was comfortable. Now it's too hot in there."

Don't you suppose some people would notice their graphics card spinning up its fan a LOT more than usual (which is usually never, unless you're playing a game).

Even CPU mining could be noticeable, as many CPU fans only spin up when the CPU is getting hot from 100% type work.

When PCs start grinding to a halt, the desktop graphics start lagging, etc. that's when people take their PC to the local PC doctor, or they wipe their machine.

BotnetPopulation = BotnetPopulation - 1


P.S. We had this discussion months ago, back in the glory days when BTC were still $18 or more each. Remember -- botnets have a monetary value for their owners. Now that a Bitcoin brings in a lousy $5, it's even less economically worth it to risk giving up an infected PC.

[P4man]

Mining peaked around August 1, 2011.
The mining graph has roughly the shape of the Bitcoin/USD price graph, but trails it by about two months. One interpretation of this is that miners hang on an average of two months after they start losing money, then quit.

A proper economic analysis of this needs to take into account both the initial cost, buying the video cards, and the running cost (electricity). Back in June, miners thought that the price was going to keep on going up, so it was seen as worth it for a very large number of new miners to come in, and they did, paying the initial cost. The price went far down, and it is now no longer worth it for new miners to enter (unless they're not paying for their electricity), but even in July it was worth it for existing miners to keep mining.

In August, however, the price kept on falling, and fell below the threshold so that the least efficient of even the existing miners were forced to shut down, which is why it's only then that the number of miners started to go down. It's not a 2 month lag, it's http://en.wikipedia.org/wiki/Hysteresis - like a thermostat only increasing or decreasing the power to the heating when the temperature falls substantially below the target. The price threshold for miners to stay mining is about 1/3 of the threshold for there to be new miners, so there is a 3x range within which the price can change without changing the number of miners, but if the price falls outside the range it drags the number of miners up and down with it (technically, it's the expectation of future high prices that brings new miners in, and that's what happened in June, but it's a good simplification in the long term).

I mostly agree, except there is a delay. Bringing new rigs online takes time. First the miner has to be convinced its worth it, and the price increase isnt a fluke,  so he may watch prices for a while before deciding to buy. Then he has to order, build, configure. It takes time. Though probably not as much as reducing hashrate; first there is the same effect, miners will look at prices for a while before being convinced its not a short term dip and decide to pull the plugs and/or sell their equipment.  Look at the for sale section, there is gazillion cards for sale there now. I bet many still have their rigs running as marginal profits may still beat marginal costs for most people, so why not use them while they wait for their cards to be sold. To be fair, I suspect most cards bought on this forum will end up mining again, but on other fora I see truckloads of AMD cards being offered to gamers from miners who threw the towel.

In the long run, particularly if bitcoin price remains less volatile, this effect will fade. Miners are learning, some the hard way, the economics behind mining.  If there is a new big bubble you might get a similar effect from newcomers though.

[SgtSpike]

Regarding botnets GPU mining on the sly --

As someone pointed out, that's very difficult as many people would wonder why their office is suddenly 10 degrees hotter than it was last week, with the same outside temperature and/or household climate control settings.

"Last week, my A/C was set at 78 and my office was comfortable. Now it's too hot in there."

Don't you suppose some people would notice their graphics card spinning up its fan a LOT more than usual (which is usually never, unless you're playing a game).

Even CPU mining could be noticeable, as many CPU fans only spin up when the CPU is getting hot from 100% type work.

When PCs start grinding to a halt, the desktop graphics start lagging, etc. that's when people take their PC to the local PC doctor, or they wipe their machine.

BotnetPopulation = BotnetPopulation - 1


P.S. We had this discussion months ago, back in the glory days when BTC were still $18 or more each. Remember -- botnets have a monetary value for their owners. Now that a Bitcoin brings in a lousy $5, it's even less economically worth it to risk giving up an infected PC.

Honestly though, I mine on two machines I use daily.  It would be very difficult for me to notice a difference.  The fan on the GPUs are still nice and quiet (they're not overclocked), and there's no noticeable difference in heat.  I have a window A/C at home set to a specific temperature, and it stays there, regardless of heat output in the room.  The other computer is contained in a room of several hundred square feet, so obviously not going to be able to tell a difference in a room of that size.

There's no lagginess on the desktop, except slight lag when scrolling an Excel document, for instance.  But not something that someone who didn't have extensive knowledge of desktops wouldn't already know.  Games run perfectly, no lag.  If the botnet was set to mine very passively, as I have my computers set, it would only lose 5% or less of the total potential production, while being nearly invisible to the user of the computer.

There are more laptops than there are desktops. There are FAR more corporate style desktops than there are gaming rigs. I would be surprised if 1/10th of the PCs out there would be remotely suitable for mining.
Scratch that, more like 1/100, as it seems even among gamers,  only 5% has DX11 capable hardware:
http://store.steampowered.com/hwsurvey

Ok, let's go with 1/100.

Say you have a botnet of 50,000 computers infected with your GPU-mining trojan.  500 effective mining machines (say, 250MH/s each on average), and 49,500 ineffective miners (say, 2.5MH/s).  That's a 250 GH/s botnet.

Now, say you have 20 friends who have botnets themselves.  5 TH/s.

I'm not saying it is true (who knows how many computers might be infected except A/V companies), but only that it is possible, and I would not be surprised if a large portion of the current mining capacities were made up of botnets.

[DeathAndTaxes]

Honestly though, I mine on two machines I use daily.  It would be very difficult for me to notice a difference.  The fan on the GPUs are still nice and quiet (they're not overclocked), and there's no noticeable difference in heat.  I have a window A/C at home set to a specific temperature, and it stays there, regardless of heat output in the room.  The other computer is contained in a room of several hundred square feet, so obviously not going to be able to tell a difference in a room of that size.

LOLZ.  What GPU.  On say a 5970 even underclocked the fan is very noticeable.  Not "oh my god I can't stand it" but no gamer is going to miss the fact that their GPU sounds like it is playing Crysis while sitting at the desktop.

Say you have a botnet of 50,000 computers infected with your GPU-mining trojan.  500 effective mining machines (say, 250MH/s each on average), and 49,500 ineffective miners (say, 2.5MH/s).  That's a 250 GH/s botnet.

Yeah @ 100% load 24/7/365 with heat blazing, fans screaming, and nobody noticing.  If you run if at lower load then it isn't going to generate anything like that.  Plus 50K computer botnet puts it in the top 3 botnets in the world.  There are 20 "friends" each with 50K botted computers (oh which also happens to have top of the line graphics cards on a significant fraction of them).

http://www.darkreading.com/security/security-management/208808174/index.html

Botnets are far more valuable for quick high value targets (like cracking password hashes, DDOS attacks, etc) not running 24/7/365 mining coins which is very easily detectable by end users.

[CanaryInTheMine]

what's postulated here regarding botnets and bitcoin is lala-land...

no-one is going to not notice GPU being in use and they will figure it out...

further more, botnet'ed computers are probably not modern systems with up to date security (unless it got disabled, yeah, sure there maybe a small number, but not constant like this)... so that leaves botnets to be full of "older" poorly attended to/neglected computers... this is not going to be a system with lots of computing power by any good measure.

so, CPU is the only thing a botnet is good for for bitcoin hashing...

so if we assume that this really is out there on a MASSIVE scale of 50K machines and if you assume 2 Megahashes per CPU (could be hidden when idle), we then have ~100,000 Mhashes/sec. or 100 Gigahashes?  is that right?

this would be hardly noticeable...

does this make sense?

[SgtSpike]

Honestly though, I mine on two machines I use daily.  It would be very difficult for me to notice a difference.  The fan on the GPUs are still nice and quiet (they're not overclocked), and there's no noticeable difference in heat.  I have a window A/C at home set to a specific temperature, and it stays there, regardless of heat output in the room.  The other computer is contained in a room of several hundred square feet, so obviously not going to be able to tell a difference in a room of that size.

LOLZ.  What GPU.  On say a 5970 even underclocked the fan is very noticeable.  Not oh my god I can't stand it but no gamer isn't going to notice their GPU sounds like it is playing some Crysis while sitting at the desktop.

Say you have a botnet of 50,000 computers infected with your GPU-mining trojan.  500 effective mining machines (say, 250MH/s each on average), and 49,500 ineffective miners (say, 2.5MH/s).  That's a 250 GH/s botnet.

Yeah @ 100% load 24/7/365 with heat blazing, fans screaming, and nobody noticing.  If you run if at lower load then it isn't going to generate anything like that.  Plus 50K computer botnet puts it in the top 3 botnets in the world.  There are 20 "friends" each with 50K botted computers (oh which also happens to have top of the line graphics cards on a significant fraction of them).

http://www.darkreading.com/security/security-management/208808174/index.html

Botnets are far more valuable for quick high value targets (like cracking password hashes, DDOS attacks, etc) not running 24/7/365 mining coins which is very easily detectable by end users.

On a 5770 and 5850.  Neither are noticeable with the fan and GPU running.

I didn't realize botnets were that small... so I guess that counts my suspicions mostly invalid then. 

[DeathAndTaxes]

I didn't realize botnets were that small... so I guess that counts my suspicions mostly invalid then. 

Well I wouldn't call 20K to 250K nodes small.  There are likely less than that mining bitcoins.  It is the rise of GPU that has provided bitcoin some level of protection to botnets.

Very rough (back of napkin numbers).


Current network hashing power is ~10TH/s.  Most of that comes from GPU.  Say average miner has 200MH/s of hashing power (now likely the average is higher but that just helps prove the point). That means bitcoin is powered by ~50,000 average GPU.   Say GPU mining didn't exist and instead everyone was CPU mining.  50,000 CPU would be enough to gain 51% of hashing power on the network.  Yeah very simplistic numbers but the reality is likely more difficult.  It is far easier to mine with 2, 3, 4 GPU than it is to mine with 2-4 CPU so the average node under a CPU only system would likely be much weaker.  Maybe even 80% weaker at which point even a 10,000 node botnet could present a real danger.

GPU caused two things:
1) made average mining hardware more powerful than average botnet hardware
2) created a compounding effect because it is relatively easy to mine with multiple GPU.

That combination means the average mining node is magnitudes more powerful than the average botnet thus requiring a staggering amount of bots to compromise bitcoin network and also makes the profit w/ subpar hardware minimal.

[mikethebodacious]

Are there other miners out there like me who are still turning a profit?  When Bitcoins were $18 a piece I was making $300/month, now that the price is down I'm still making $90/month so I see no reason to quit.  I initially bought more efficient hardware (80 Plus Gold PSUs and Radeon 5000 series cards) so my watt consumption is rather low.  I am on the fence about building another rig at this time, I want to get another 5970 since the heat output will be negligible in the winter (just crack open the nearby door) but don't want to deal with that extra heat come next summer.  The 7970s coming out have made me reconsider investing again so soon.  My rigs are paid off at this point I'm just waiting to reinvest.  I definitely will not be buying any more 6950s since their hash/watt ratio is nowhere near 5770s or 5970s so hopefully the 7000 series cards will be far more energy efficient.

[DeathAndTaxes]

Are there other miners out there like me who are still turning a profit?  When Bitcoins were $18 a piece I was making $300/month, now that the price is down I'm still making $90/month so I see no reason to quit.  I initially bought more efficient hardware (80 Plus Gold PSUs and Radeon 5000 series cards) so my watt consumption is rather low.  I am on the fence about building another rig at this time, I want to get another 5970 since the heat output will be negligible in the winter (just crack open the nearby door) but don't want to deal with that extra heat come next summer.  The 7970s coming out have made me reconsider investing again so soon.  My rigs are paid off at this point I'm just waiting to reinvest.  I definitely will not be buying any more 6950s since their hash/watt ratio is nowhere near 5770s or 5970s so hopefully the 7000 series cards will be far more energy efficient.

I think in the medium term nothing can touch the 5970s and that includes the futue 7xxx series.  The 7xxx series will likely close to double performance per watt and performance per $ however the prices will be back to full retail cutting into the value.

So while a 7990 will likely be in the ballpark of 2x the performance of a 5970 (or 6990) it will also be $750 to $800 making the hash per $$$ roughly the same or maybe a little worse. I bought some 5970s @ $400 and some used ones on ebay @ ~$350 simply because that is an amazing value even considering the potential for 7xxx series.  After the 7990 hits I imagine many gamers will drop their 5970s on ebay so there should be a good market for me to cheaply expand hardware. 

I am actually slowly getting rid of my 6950s and 6970s because the 5970s beats them in every category (density, price, efficiency).  My goal is to consolidate my hardware (currently 8 rigs both open and closed cases with 3-6 cards of various types) into 6 standardized open frame rigs of 4x 5970s ea in the garage plus my watercooled workstation w/ 3x 5970 in the office.  That should give me ~19 GH/s (maybe an even 20GH/s if I can tweak a little more out of em) at very low cost per MH and decent MH/kWh (~2MH/kWh).

Now on price per watt level the 7xxx series is going to rule however my electrical costs are relatively low (about $0.09 per kWh) thus my gamble is that lower capital cost is worth more than lower power costs (at least initially).  At current difficulty a block costs me $90 in electrical costs.  The price / difficult ratio would have to fall 64% from current levels (price falls or difficult rises) before I am below break even.

[mikethebodacious]

I think nothing can touch the 5970s and that includes the futue 7xxx series.  The 7xxx series will likely close to double performance per watt and performance per $ however the prices will be back to full retail.

I agree completely.  I hated my 5970 at first but seeing the hash/watt performance nothing else comes close, for long-term mining these are hands down the best card.  The fan was loud for the first month but after it burned in I lowered my fan speed 10% and it's not much louder than a table fan on high.  I wish I had money today to buy some on Newegg with the $100 off promo, hopefully they extend it one more week and I can nab another one.  The 7xxx series will be worth it in the long run if their hash/watt performance is close to the estimates so far.  Looking forward to those cards

[racerguy]

I think nothing can touch the 5970s and that includes the futue 7xxx series.  The 7xxx series will likely close to double performance per watt and performance per $ however the prices will be back to full retail.

I agree completely.  I hated my 5970 at first but seeing the hash/watt performance nothing else comes close, for long-term mining these are hands down the best card.  The fan was loud for the first month but after it burned in I lowered my fan speed 10% and it's not much louder than a table fan on high.  I wish I had money today to buy some on Newegg with the $100 off promo, hopefully they extend it one more week and I can nab another one.  The 7xxx series will be worth it in the long run if their hash/watt performance is close to the estimates so far.  Looking forward to those cards

you can get similar perf/watt with normal cards by not over clocking them or undervolting and slightly underclocking (which is how the 5970 does its magic).

[DeathAndTaxes]

True but then 5970 is just even better in terms of MHash/$ by comparison.

What makes the 5970 nice is the combination of BOTH high MHashes/$ AND MHashes/Watt.

[bronan]

Seriously i do not think a big portion of miners really left, but am certain many are waiting to bitcoin become profitable for them again.
I am fairly new to this but when i saw what my 5870 brings in 10 minutes i already knew it was not going to be impressive
The friend who asked me to see this, said to me that you should run it about a day or so to get a block.
Which would give me some higher gain.... and according to him he got 1 block about every day giving him something near 1.xxxx btc
After 40+ hours running in a pool i made the super big amount of 0,21 BTC and non block ever showed up.
I only see the top guys scoring up to hundred or more blocks

But for beginners like me its impossible to get even a tiny profit in the current state, because even while many miners are inactive or left the difficulty is still darn high to give people like me a change

But now comes the fun with the current price of electricity in our country we pay that company a fortune.
And for what to receive 1/10 the value i spend and probably even a lot less
So i constant am amazed people dare to say that mining is profitable, except the ones living in a country where the cost of electricity is still low (usa)
I am certain that non of the people in england/germany/netherlands/belgium are loosing money when mining if they would do it
Ofcourse the old school miners who made a fortune at the start of the bitcoin did, or the ones who setup a pool benefit probably pretty well here.
Others who have tons of money probably can risk trading, but again i am not one of them who can do that
So only those who do not pay the electricity bill themselfs can make some profit but when mom/dad finds out they won't be happy.