bitcoin network the top supercomputer in the world?

[speckman]

I'm pretty sure (but correct me if I'm wrong) that the bitcoin hashing network beats out the top supercomputer in Japan, and in China (which maybe has a 3x Japan's). 

Japan's has 8 petaflops.  Roughly, the mining cards have about 300 peta-integer-ops, and I think 30 petaflops.  Just rough calculations based off of what I remember from ATI's specs, and some pretty dubious assumptions about what cards are running and how many integer ops ati cards actually have versus flops.  And since it keeps growing like crazy, if it isn't now, it's close and will be soon.

[1715001607]

1715001607

[1715001607]

1715001607

[1715001607]

1715001607

[Mortox]

Its been posted before, but its very hard to make accurate comparisons between supercomputers and bitcoin mining.  They really aren't doing the same things.

[speckman]

Oh, I'm sure it's been posted before.    Just writing to write.

They're definitely not doing the same thing, but the gpus are going full-out, doing integer ops.  Supercomputers probably do floating ops because they're doing simulations and whatnot, but we know the rough floating/integer ratios for ATI cards.  So they're all networked together, crunching away, getting cooled like crazy, doing neverending work.  And quite possibly, the whole thing, on average or by sum, paid for itself. 

So it's the cheapest (ie. free) and fastest supercomputer in the world.  Way to go, guys!

[kjj]

Well, linpack benchmarks all measure in FLOPS, and the bitcoin network is still at zero FLOPS, so by the accepted benchmark, we lose.

However, we pwn the entire Top 500 list if you measure MIPS.

Then again, we have no spare capacity, so in a race to do any sort of computation, we will lose.

[thegreatall]

Keep in mind that lot of people are using graphics cards to do mining, and graphics cards are very limited in what they can process. In other words, if we where able to chain all machines together and solve a problem i'd likely be able to produce a super thousand frame per second 3d movie (like what Pixar makes), but would be terribly inefficient at somethings like floating point, string concat, or generally anything not having to do with polygons and shading.

[squirrall3]

Combined total of Top500 supercomputers is 43673 Tflops
http://www.top500.org/lists/2010/11/performance_development

 Bitcoin Grid GPU/CPU estimate 112989 Tflops
http://www.bitcoinwatch.com/

[eramus]

Considering the way the bitcoin network works, I would assume it pales in comparison to todays supercomputer. The number of flops may be substantially higher, but they are not working together as a single unit to complete a single task. What we have with the bitcoin network is a ton of machines (pools or individuals) doing the same job separately. So, in reality the real comparison would be to pick out the single strongest machine (or pool) to make the comparison. I would be very surprised if a supercomputer did not come out on top in that instance.

[Confucius]

Bitcoin network is just that - a network. Its like saying the internet is the largest super computer. I would use the term most powerful 'computing network'. So Bitcoins is a pretty powerful computing network in one regard, while the internet is and always should be the most powerful computing network.
If everyone on the internet was running software that collectively does some calculations, it would be pretty fast at doing so. Folding@Home is an example.

[Gabi]

The supercomputers in top500 aren't made of GPUs... sure, we have enough GPUs to have much more FLOPs than them but can we do what THEY do faster? No.

Also the whole network hashrate is worth like 10/15 millions of $, the price to buy enough graphic cards to make 9000Ghash. And even 15 millions isn't so much compared to even only one of these supercomputers!

[Samantha2011]

How many FPS could we get if we used the Bitcoin network to run Crysis? 

[cothoms]

It would obviously vary, depending on whether or not we were using gold-plated monster cables, Samantha =)

[Samantha2011]

It would obviously vary, depending on whether or not we were using gold-plated monster cables, Samantha =)

Of course! I always use the highest quality gold plated $200 HDMI cables. 

[btcbetnet]

I´m really interested in what this powerfull mashine really calculate.

[JoelKatz]

I´m really interested in what this powerfull mashine really calculate.

It piles more and more computational effort on top of invalidating past transactions.

[speckman]

squirral: wow!  Wow.

eramus:  Sure all of the miners are doing independent tasks, but the whole process is a brute force search for some really unlikely things.  So they all have to work independently, sort of, it's the nature of the task.  It's still a giant task.

[coinage]

I´m really interested in what this powerfull mashine really calculate.

Sadly, all the calculations are intentionally pointless.  It's "make work" for computers.  Basically a lot of random numbers are created then thrown away.  It's fishing the ocean of numbers for the smallest fish (hash) still left to find.

I'm very pro-bitcoin, but disappointed that its visionary developers haven't yet come up with a way to pair an actually useful computation with the fantastic amount of hardware and electricity we've amassed.

The project itself is extraordinarily valuable, however.  The proof-of-work concept that powers and regulates it has in fact brought about a practical worldwide cryptocurrency and that is an astonishing feat.

And in the future when mining becomes unprofitable for power-hungry GPUs and shifts to optimized, low-power devices, we may see the system self-adjust the frenzy downwards to a much less wasteful level.  Or we may see a transition into a useful substitute for hashing which yields profits while simultaneously maintaining the reliable flow of BTC.  Otherwise the costs of all the electricity and periodic hardware replacement may make bitcoin more expensive than some other alternative, which might then overtake it.

[Globz]

Nope!

[kjj]

I´m really interested in what this powerfull mashine really calculate.

Sadly, all the calculations are intentionally pointless.  It's "make work" for computers.  Basically a lot of random numbers are created then thrown away.  It's fishing the ocean of numbers for the smallest fish (hash) still left to find.

I'm very pro-bitcoin, but disappointed that its visionary developers haven't yet come up with a way to pair an actually useful computation with the fantastic amount of hardware and electricity we've amassed.

The project itself is extraordinarily valuable, however.  The proof-of-work concept that powers and regulates it has in fact brought about a practical worldwide cryptocurrency and that is an astonishing feat.

And in the future when mining becomes unprofitable for power-hungry GPUs and shifts to optimized, low-power devices, we may see the system self-adjust the frenzy downwards to a much less wasteful level.  Or we may see a transition into a useful substitute for hashing which yields profits while simultaneously maintaining the reliable flow of BTC.  Otherwise the costs of all the electricity and periodic hardware replacement may make bitcoin more expensive than some other alternative, which might then overtake it.

What could be more important than proving the security of a global money system?  Finding unlikely hashes can't be faked.  Any other system would allow the possibility of cheating.

[speckman]

Yeah, it is disappointing that nothing else is happening, like proteins being folded, cancer cured.  

I'm thinking though that more namecoin like projects will arise.  (I'm not too hip so I don't know about them yet.)  I'd like to see some way of doing the folding/seti@home type stuff with the bitcoin network (or rather people moving to it from bitcoin), a way in which people get paid for their work, but I'm not sure why that currency would have value.  Saw mention of a possibility of piggybacking onto the bitcoin blockchain for security? for other *coins. 

Like imagine there's an AI network, crunching away at finding better AI (sounds fishy but I'm just trying to summarize instead of spending 4 pages explaining ).  Maybe if the miners are getting paid in shares, like stock, that can be traded for btc.  Like this AI company is going to use the power of the network to make really well-evolved stuff, and so is going to pay them to do it (in future dividends per share?), and the better it is, the better the company will be in the marketplace.  Possibly some small percentage of the power of that network would go to finding hashes for their currency or distributing the payments a la bitcoin, but it would be a very small difficulty, and hopefully could somehow piggyback on something else just for the security.

But the point here is that the bitcoin network is amazingly powerful, and it could very well funnel a lot of that power into related projects, just because everyone's sort of actively involved--more likely to see it and go there, being on bitcoin, than if they were just sitting in their quiet house, watching hulu, oblivious.  Which is where I was a month ago, with no hashing power.    Now I have a gig.  And I approximately multiplied my total instructions per second in this house by 1000.  And possibly the decibels. 

[IceyEC]

Thank you, Skynet (speckman)