Bitcoin Forum

we are 25 times or so as fast as folding@home. By far the most powerful distributed computing network there is. Maybe we should claim the world record.

edit (2012-01-13): link updated:

http://www.guinnessworldrecords.com/records-5000/most-powerful-distributed-computing-network/

add your comment that the records as stated is wrong and that bitcoin is faster!


technical thread on how to estimate the calculation power of the bitcoin network:

https://bitcointalk.org/index.php?topic=38064

http://bitcoinwatch.com says we are at: 149796 tera flop/s (149 Peta flop/s?) not at 25 petaflop/s as you claim.

EDIT:

Iam correct on this right?
Damn thats just so much faster then the fastest super computer out there (K-computer - 8.1 Petaflop/s)

We certainly seem to deserve that spot :)

 No doubts, in 2 years we will be there  :D...

well let us file for it then . who should do that ? Satoshi or andreson ? umm maybe me ! :D

I'll file as long as it's free. which, I doubt.

Instead of doubting just go and read
http://guinnessworldrecords.com/member/faqs.aspx
http://guinnessworldrecords.com/member/how_to_become_a_record_breaker.aspx
http://guinnessworldrecords.com/member/is_it_a_record.aspx

FAQ #12

I know a lawyer in New York who's probably already filed for the record.  :-\

great idea, I'd love to see bitcoin immortalized in print in the Guinness Book

This would be some great publicity.

I love the guinness book of world records!

They also make a really great beer.   ;D

This would give Bitcoin some credibility in the public's eyes. Definitely a worthwhile endeavor.

Yeah, but Guiness is a little weak for my tastes ;>

Arrogant Bastard World Records ??

a world record would give bitcoin a push! normal people will realize how big bitcoin already is!

who's going to set it?

Guiness (beer) does not travel. You cannot pass a judgement until you tried it in Dublin.


As for the record claim, it is an awesome idea and it must be done. Free good publicity is even better than free bad publicity.

i'll try it.
but i live in switzerland, it could take weeks until it's accepted...

would you share the cost of it with me?

You're talking about beer, right?

yes guinness!


and for the persons who are interested in bitcoins not beer:

a fast claim wil cost about 30 BTC  (3days instead of 4-6 weeks)

shall i do this? will you help me pay?

just do free filing, there is no rush

This is a very good idea; has anyone made the effort to get in contact with them? The community is obviously for it. Many people have invested a lot of money, and are proud of their gear. They wouldn't mind showing it off to the world!

I'm not sure the fast track is especially necessary - I personally think slow and steady growth is much healthier for Bitcoin so I don't think we're in a need to rush this. Besides, we're probably not at risk of being overtaken by anything in the next 4-6 weeks :)

Having said that, if the community deems the fast track to be the better way then I'd be happy to contribute 1 BTC to the cost.

Yes i'm filling out the claim right now!

hope well get the record!

on bitcoinwatch its 150petaflops!

Anti-procrastination! I like your spirit.

How can i proof the Flops?
any idea?

http://bitcoinwatch.com/

do not use bitcoinwatch numbers for this

the best way would be to contact user Raulo who is kind of resident expert on this and ask for his advise

What is more accurate? Sorry if I misled.

YES.


That is all.

If we take current hashing power as 11857 Ghps
and make an assumption that it is all generated by the most efficient card 5970 which is rated at 4.4 TFLOPS (out of top of my head) this card could also do 0.75 Ghps.

Therefore, we need 11857 / 0.75 = 15809 cards which at best could be together rated at 4.4 * 7905 = 70 PFLOPS .  

These calculations are already extremely optimistic IMO.

Where bitcoinwatch takes that 150 PFLOPS I am not sure. A 5970 must do 10 TFLOPS for their number to be correct.

EDIT: messed up hash rate of 5970 initially, fixed now, sorry. Actually bitcoinwatch is not that far from my number and assuming that some cards a less efficient than 5970, their number could be actually quite accurate.

I would point to the graphs at http://bitcoin.sipa.be/speed-lin-2k.png (http://bitcoin.sipa.be/speed-lin-2k.png) and refer to the hashing power of the most often used (and most efficient) GPU, the 5970. The hashrate of around 12000 GHash/s would need about 15000 ATI 5970 GPUs (~800MHash/s according to the wiki (https://en.bitcoin.it/wiki/Mining_hardware_comparison)).
Since we're not doing FLOPs when hashing we can only estimate what our FLOP/s rating would be:

Depending an how you measure the power of a 5970, the maximum theoretical computational capacity of a 5970 is 4640 GFLOP/s single precision or 928 GFLOP/s double precision. That means we are at about 13.92 PFLOP/s double precision (again: theoretical maximum).

The current entry for the record-holding Folding@home project is not very precise either, so I think you should take the number of GPUs as a starting reference (something like: "well over 30000 GPU cores, resulting in about 14 quadrillion floating point operations per second").

EDIT: Vladimir, I think you're off by a factor of 2 with the hashing power of a single 5970

EDIT 2: I think FLOP/s ratings usually refer to double precision, so we should probably go with that as well, even if the numbers don't look quite that impressive then

yep, you are right on both my 2x mistake and usage of double precision

I just took a look at the foolding@home client stats (http://fah-web.stanford.edu/cgi-bin/main.py?qtype=osstats). They seem to have about 20000 GPUs (about half ATI) and about 20000 Playstation 3. They rate GPUs at about 1500 GFLOP/s and a Playstation 3 at about 600 (native) GFLOP/s.

If this data is correct then our lead is probably thinner than we thought...

They must mean single precision GFLOPS there than.

Yes you are right - it seems they are using single precision operations.

From the page linked above:

I honestly don't know what's the best way to compare Bitcoin with Folding@home, also because it is unsure how their definition of "active" translates into sustained processing power:

only important that we lead not how much!
how can i proof?
i think i cant calculate them this:

Well, the "proof" is the blockchain, but I think a well formulated claim with some links to further information is sufficient.
As I said in my previous posts, you could use a wording similar to the current entry in your proposal:

"On 16 July 2011 Bitcoin, a voluntary peer-to-peer computing project to create the worlds first decentralized digital currency, achieved a computing power equivalent to over 14 petaFLOPS (14 quadrillion floating point operations per second). The project was initiated by a cryptographer named Satoshi Nakamoto in 2009. Since then, a growing number of Bitcoin users around the world combine the processing power of their computer's graphic cards to create and secure those new digital tokens of value. Counterfeiting Bitcoins would require an adversary to have a much higher computing power than the whole Bitcoin network combined, which makes them practically forgery-proof."

For further reference, include links to the hashrate graphs (http://bitcoin.sipa.be), the bitcoin.org (http://bitcoin.org) project page, possibly the weusecoins-video (http://weusecoins.com), the introductory page from the wiki (https://en.bitcoin.it/wiki/How_bitcoin_works#Bitcoin_mining) and the link to the original paper (http://www.bitcoin.org/bitcoin.pdf) from Satoshi. I don't think they will need much more of the technical details.

+ 1

This should be good enough. Guinness probably has fairly competent researchers or/and access to experts to verify the claim.

there has to be answerd:
Who is attempting this?
How will they do it?
How is it measured, what is the record based on?
Why are you doing this?

Maybe encourage bitcoinwatch to help you file. It's possible that they could get some publicity as well. Surely they get their numbers from somewhere

its done i've sent the claim... now we have to wait

i'm not the owner of the bitcoin system... hope i dont get in trouble because of it (if satoshi arrives: "he it's my record")
i hope they're enough intelligent, that they dont give the record to me. its for the bitcoin community!

it made me think, too. "Most powerful distributed computing network in the world." - Yummy!

the numbers on bitcoinwatch are based on assumptions in this thread (and the blockchain I guess):

http://forum.bitcoin.org/index.php?topic=4689.0

stole my joke!   

Sent ya some for your work.

Thank you..

Seems to me that the bitcoin network is doing integer operations, not floating point.  So what's the relevance of FLOPS?

We're well aware of that (as you surely noticed while reading this thread). The FLOPs measure is not applicable to hashing per se, but you can estimate the FLOPs potential of the Bitcoin network based on the approximate hardware making up the hashing power.

Since the folding@home people have their own way of measuring their computational power which makes it hard to compare it with the Bitcoin network it's probably the best we can do.

I wrote an email to Nils Schneider.
he said:
"The fundamental problem is, that bitcoin does not use floating point
operations. What you can do is: Pretend that the same hardware was build
to do FLOPs instead of INTOPs and then you would get about 153 PetaFLOPs."

it's larger than we thought. I will tell this argument guinness world record if we dont get the record the first time...
if they negatiate too, we can make an own record with hashs!

I believe whatever calculation method they will approve Bitcoin will be the highest PFLOP single project present on earth.

I think I will file (too) if no one here said clearly he filed for it.

i have already requested the record, i'll tell you everything, if i've got news.
but the free request takes 4-6 weeks until first response...

That would be good if there was some way of getting a rough idea of the hardware making up the hashing power.  Do we have anything better than guessing at the moment?  I don't think any peer on the network even reports that it is hashing unless it solves a block, let alone having to identify the hardware it is using.

I think the only way we can estimate the total hashrate of the network as a whole is to look at the rate that blocks are solved relative to the current difficulty.  Is that right?

Do we have any idea what percentage of the network's hashing power is made up of ASICs which have no floating point capability at all?

Got news from Guinnes world records:

MOST POWERFUL DISTRIBUTED COMPUTING NETWORK
DEFINITION OF RECORD
This is for the most powerful distributed computing network.
It must be measured in petaFLOPS and floating point operations per second.

GUIDELINES FOR ‘MOST POWERFUL DISTRIBUTED COMPUTING NETWORK’
1. For the purposes of this record, a distributed computing network consist of
clients and servers connected in such a way that any system can potentially
communicate with any other system.

2. The capacity of the distributed computing network must be tested using
appropriate means.

3. The exact measurements of the distributed computing network must be given in
petaFLOPS and floating point operations per second.

4. An expert witness, independent of the computing network, must attest to the
computing power.

5. Two (2) independent witnesses must attest to the computing power.

6. Any other relevant information must be included in the cover letter or evidence
package.

GENERAL ‘MOST POWERFUL DISTRIBUTED COMPUTING’ NETWORK
GUIDELINES

§ The name of the organisation, company or person(s) making the attempt must
be given, along with the date and place.

§ Failure to include the required documentation will ultimately delay the outcome
of your claim or lead to its rejection.


-> How can i fix or demonstrate them all these points?

Is there even a valid answer for that with somthing decentralized like Bitcoin?

"Bitcoin, all the time, all over the planet"  ;D

Organisation:Bitcoin.org
People: Inhabitants all over the world?

How did the folding@home people fill that in then?

This would be awesome. Please keep us updated.


I wonder what would be considered "independent". I'd guess anyone who's not mining.

Also, "expert" is another vague one.

Getting this would be awesome, and would immortalize bitcoin.


The mining nodes do this by uploading and downloading blocks to/from each other. Any transaction that one node receives will be communicated to every other node. Any node that finds a block meeting the proof of work conditions will have their block communicated to every other node.


The average time it takes to find blocks at a particular level of difficulty measures the capacity.


I'm not sure how this can be done.



I'm sure we can find two expert witnesses.


Gavin would be the most official person in the bitcoin project. "Lead developer for the bitcoin open source project".

heh... 

The speed of the bitcoin network, measured in FLOPS, is exactly zero.

Can we just say the miners are using floating point operations involving 1.0000000, 2.000000, 3.0000000 and so on? :D

That is not exactly true. Most bitcoin software, including bitcoincharts.com does some incidental Floating-point calculations for display purposes.

Well, that depends on what you mean by the "bitcoin network" - most of the underlying hardware is quite capable of floating point operations so we can very well assign a FLOPS rating to the Bitcoin network (meaning all the mining hardware).

The problem is of course that we don't know exactly which kind of hardware makes up the whole hashing power. The other problem is that since we don't use floating point arithmetic, we could only quote some theoretical performance values anyway.

I'd say the best thing would be to point out that we're just doing IntOPS and then offer a transparent calculation for a FLOPS estimation: Assuming that the vast majority of the calculations is performed by ATI GPUs (I know that's debatable...), one hash requires about 3800 IntOPS with current OpenCL kernels - so with a hashing power of 13000 GHash/s that would equate to about 50 PIntOPS.

FLOPS-ratings of GPUs are usually 2*IntOPS, so theoretically about 100 PFLOPS peak. Of course there are other estimations but you could as well offer various different estimations if they insist on a FLOPS rating.

Just make the calculations transparent and quote sources. The ultimate proof of the computational power is of course condensed in the last block of the blockchain and the way it was created (ie. the Bitcoin system). Any mathematician can objectively estimate the current computational power of the network from that alone - albeit just INTOPs.

Maybe you could start a wiki page with all the calculations and we can together collect references and refine the arguments.

I have a good contact to a german publicly certified computer and network assessor (in german: öffentlich bestellter und vereidigter EDV-Sachverständiger). I will ask him if he has time to attest for bitcoin. Will there be a bounty for this?

That's not even relevant. It says "[...] clients and servers [...] can potentially communicate with any other system.". "potentially" means - they don't have to, they just have to be interconnected. And, being connected to the internet, that's a given.

Interesting link - Folding@home shows as 6PFLOPS

Individual contributions (an indication) at http://boinc.berkeley.edu/chart_list.php so the top BOINC guys are well behind some of the bigger mining people - but then, BOINC doesn't pay any bills.
Sample single project Milkyway is running around 400TFLOPS http://milkyway.cs.rpi.edu/milkyway/server_status.php
Combined http://boincstats.com/stats/project_graph.php?pr=bo says 5 PFLOPS, and is also cited at http://en.wikipedia.org/wiki/FLOPS running at 5.3PF.

As a post-it-note calculation, I have about 20TFlops for 3Ghash out of 15,000Ghash.  I can can turn in about 2M RAC out of the 1B RAC over in boinc as a comparison.  Implies BTC network is running around 100Peta flops +/- a lot.

Aren't they way more centralized than Bitcoin?

Some more interesting stats...

http://boinc.netsoft-online.com/e107_plugins/boinc/bp.php?project=1

I looks like since may all the BOINC projects have been dropping.  Folding@home peaked at around 8 and is now around 4. Coincided pretty much with the bitcoin price bubble and the Silk Road publicity.

technical discussion here on how to best prove the record:
https://bitcointalk.org/index.php?topic=38064 (https://bitcointalk.org/index.php?topic=38064)

this is not going forward.

I suggest we figure out how to prove the bitcoin network performance in a community effort and hand the information to any expert that will review it and hand it in. Please note the other thread in my post above.

Bitcoin could well need some positive publicity at the moment.

Agreed - I suggest to start a wiki page with all the information and references that were posted so far and try to form it into a conclusive and well written article. Once we have that, we can probably find some experts within the community for a review.

And we better do this quick as the network is shrinking.

A page on the bitcoin wiki for this purpose sure would be very nice. I am not sure if we can start a movement to use it, though. Even in the technical thread about this topic on this forum the wild discussion has not yet started...   (https://bitcointalk.org/index.php?topic=38064)

The Guiness Book of world records lost all credibility to me when they started allowing geographically-distributed records such as "x people walking at the same time." I suppose any Folding@home or Bitcoin record would be similar. The Internet has acces to more FLOPS than any individual project :).

My point is records for doing things everybody else is many other people are doing anyway don't really mean much.

I think the record would convince a lot of people that bitcoin is big. at least it would get us publicity.

what about this new strategy:

just claim the record until someone can prove the opposite.

by my estimate we are quite a bit faster than all the competitors, so it is pretty obvious to me: https://bitcointalk.org/index.php?topic=38064.new#new

Can you ask them about IntOPs?

It is like this: bitcoin has a box of oranges and folding@home has a box of bananas. We could only theoretically exchange our oranges for bananas and have much more bananas than they have. Still, we actually have more than ten times more fruits than they have.

But they only accept bananas. Our oranges are TOTALLY useless.

Only an idiot would measure computing power only in flops

as far as I know we have not yet asked them about anything else. we need to explain the problem and convince them to add an IntOPs category and a general category where it is ok to use voodoo converted OPs.

I'd really like this thing to take off...  Bitcoin for World Record!

updated the original post with correct link to the Guinness website... feel free to add your comments there

suggested this issue in the bitcoin foundation thread:
https://bitcointalk.org/index.php?topic=49841.msg689941#msg689941

Watching

added a claim to wikipedia - who want's to bet how many minutes it will stick?  :D

http://en.wikipedia.org/w/index.php?title=Bitcoin&diff=prev&oldid=476478475

That is a good point: "The Bitcoin network is by far the fastest distributed computing network in the world. <ref>{{cite web |title=Block 0 - Bitcoin mining network speed estimate |url=https://bitcointalk.org/index.php?topic=38064 }}</ref"

However, if you want to keep the Bitcoin article in a professional shape, you should avoid vague definitions, such as "by far". Numbers should be given that prove your position.

I don't think it should be noted on the top part of the article, either, since it is not a defining characteristic of Bitcoin. It would look better in a section about mining, related to P2Pool and other pools.

Finally, a reference to a forum is never trustworthy for Wikipedia, and any editor is authorized to remove it, if that is the case.

I agree most newbies get their first impression about Bitcoin from Wikipedia. So it should always be kept as professional as possible.

An example of a well written article about Bitcoin is in the German Wikipedia: http://de.wikipedia.org/wiki/Bitcoin
In section 5.4. Rechenaufwand beim Mining, it explains the computing effort in mining, and how it compares to supercomputers around the world.

I think it shows the significance of bitcoin so for me it would be ok in the top part. Of course I am aware it will be removed soon. Thanks for pointing me to the German article, looks decent and more in depth than the English version. A mining section is really missing... who volunteers to translate?

Added another bitcoin comment!

So what would "where" be? The world? ^^


Added a comment btw.

Earth, to be more specific.

grrr, it got deleted.   :'(

maybe the milken institute helps with this SIGNIFICANT publication: http://www.milkeninstitute.org/publications/review/2012_1/22-31MR53.pdf   p.26, right paragraph:


I wonder where they took that number from   :D

OK, I got it to stick for now.   ;D

http://en.wikipedia.org/wiki/Bitcoin

"The Bitcoin network is by far the fastest distributed computing network in the world."

With ASICS I'd say we went on to a completely new level so "the fastest distributing computing network" has become pointless. What about changing it to: "The most expensive distributing computing network"?  Can we find a quote for that somewhere?

If there is some prize, it should be used too buy bitcoins for that, and destroy it after that so evry bitcoin user will gain proportional value to his part of hoarded bitcoins. What do you think about that?


BTW if it would be 1M$ it should in long term increase about ~ 1% (1/110) for now.

I think in two years we'll have several guiness records...not just this one.

Has anyone mailed the gays at Guiness World Records yet?

Fun fact, with ASIC the FLOPs capabilities of the bitcoin network will be exactly 0

Some of us will vehemently keep mining with GPUs  >:(

While a world record sounds great, much of the records listed in the Guinness World Records book are not of high quality. This book and it's records have become more and more of a joke during the last years. (imo)