embedded bitcoin mining

[paulie_w]

have you guys thought about this?

every point of sale system out there, that sits idle 99% of the day...

the next world of warcraft...

or how about the next version of microsoft windows (steve balmer, sweating on stage: "Uhhh we have a surprise for you all, the next version of windows is... free... yeah, and it comes with a free high-end ATI graphics card, and we'll also pay your electricity bill. we figured fuck it, we have the money...")

huge opportunities out there for anyone with distribution capability to totally own the network and mine the shit out of some coin.

when do you think we will see this kind of thing start to happen?

[1714011285]

1714011285

[1714011285]

1714011285

[1714011285]

1714011285

[imperi]

Bitcoin mining will not always be so profitable.

[paulie_w]

but my point is... with distribution all you have to do is release a 'hotfix' or patch, and you could theoretically be backgrond-mining across more than a million computers.

[Mapatti]

but my point is... with distribution all you have to do is release a 'hotfix' or patch, and you could theoretically be backgrond-mining across more than a million computers.

It would be little odd when computer is idling but GPU is at 99%.

[mellowhead]

Could be dangerous if run on machines without proper cooling. And a mass distribution is highly likely to come across machines like that.

[paulie_w]

is there any rule that says you have to use 99% of the gpu/cpu power? why can't it be some lighter process (but made up for by being so distributed)?

[mellowhead]

I would think it's entirely possible to set it up for say 50-75% usage. The guiminer application already has that switch built in with the -s modifier. A user would be unlikely to notice any difference in standard desktop task performance. An unsuspecting user probably wouldn't notice anything unless they started running a graphics intense application or had a usage monitor like I have on my desktop.

Anyone able to sqeeze this into the next Windows update? 

[Patheos]

I work at an office building that has over 300 computers running 24/7.  I was thinking the other day that I could install  miners on all of them during the hours nobody is working since we leave them all on.  Too bad CPU mining is pretty much obsolete or I would look into it.

[bitbang3r]

As others have pointed out, pure CPU-based mining, even with uber high-end PCs standing alone, is largely obsolete. Looking at the next step down, a high-end Android phone with 1GHz ARM CPU is approximately equal in computing power to a 500MHz Pentium III (give or take, but it's a fairly reasonable approximation). A gigahertz of RISC is worth a lot less than a gigahertz of CISC. The advantage of RISC is that RISC (in theory) can be scaled more easily to run at higher speeds than CISC. But when comparing a RISC to a CISC CPU at the same nominal speed, and specifically an ARM to a non-Netburst Intel-Architecture CPU, the IA32 CPU is going to stomp all over it.

OK, now that I've established that perspective, consider further that most POS hardware has -- at best -- an ARM CPU running in the neighborhood of 200MHz, and there's a LOT of hardware out there running on m68k Coldfire & Dragonball cores that are nominally ~40MHz, but roughly comparable in performance to the same 100-200MHz ARM cores (m68k is CISC... though later Coldfire variants kind of muddy the equation).

Anyway, the point is, in order to meaningfully make use of POS-type and other embedded hardware, you're absolutely going to have to treat them like army ants and implement it as a massively-distributed system whereby you have "controllers" that orchestrate the individual computing actions of the individual devices, and do it in a way where the communications overhead doesn't end up neutralizing the actual work done by each individual worker. A good analogy is to look at programming the PS3 cell architecture. In fact, I'm surprised the PS3 hasn't gotten more attention, because a mountain of PS3s would probably be the most cost-effective, Sony-subsidized computing array you could buy if a Bitcoin mining app were available for it.

Either way, any application that targets anything less than a high-end Android phone is probably a waste of time. And anything that targets Android phones that are actually being used as PHONES is an equal waste of time, because a useful bitcoin mining app would nuke most batteries within an hour or two. On the other hand, an app that's intended to make use of hundreds of OLD Android phones that are connected to chargers, networked via wi-fi, and would otherwise be sitting unloved and unused in a drawer somewhere might have interesting possibilities... especially if you have physical access to a storage room with crates and crates of such phones.

[Sukrim]

In fact, I'm surprised the PS3 hasn't gotten more attention, because a mountain of PS3s would probably be the most cost-effective, Sony-subsidized computing array you could buy if a Bitcoin mining app were available for it.

It isn't, there are PS3 miners already available.

If you're a MH/s per Watt fan, there's nearly no way around FPGAs or ASICs derived from these.