Bitcoin Forum

Bitcoin => Bitcoin Discussion => Topic started by: drrussellshane on June 13, 2013, 05:54:41 PM



Title: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: drrussellshane on June 13, 2013, 05:54:41 PM
Well?

Quote
    OSTP’s Associate Director for National Security and International Affairs Dr. Patricia Falcone provided keynote remarks yesterday at a ribbon-cutting ceremony celebrating the opening of the US Army Research Laboratory’s (ARL) new supercomputing center at Aberdeen Proving Ground in northern Maryland...

    The new ARL Supercomputing Center—containing two new IBM iDataPlex computers with the capacity to perform 50,000 trillion floating point operations per second, or 50 petaflops—will provide state-of-the art high performance computing capabilities as well as extraordinary capacities in advanced high-speed networking and data analysis, providing unprecedented benefits to the Army, the Department of Defense, and the Nation as a whole.

http://www.weeklystandard.com/blogs/wh-celebrates-new-high-performance-computing-center-opening_735141.html


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: Gabi on June 13, 2013, 06:06:35 PM
Petaflops are totally useless for mining  ;D


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: Come-from-Beyond on June 13, 2013, 06:21:53 PM
Petaflops are totally useless for mining  ;D

I wouldn't say totally...


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: aceking on June 13, 2013, 06:29:17 PM
If the US government wants to attack to bitcoin has lot ways to do it efficiently.


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: Gabi on June 13, 2013, 06:37:07 PM
Petaflops are totally useless for mining  ;D

I wouldn't say totally...
How many floating operations are required for sha hashing?


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: Come-from-Beyond on June 13, 2013, 07:50:23 PM
Petaflops are totally useless for mining  ;D

I wouldn't say totally...
How many floating operations are required for sha hashing?

No idea. Why?

EDIT: I googled and found that u need only 3 floating-point operations for SHA-256 hashing - https://blogs.oracle.com/DanX/entry/sparc_t4_digest_and_crypto.


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: pof on June 13, 2013, 08:03:00 PM
http://bitcoincharts.com/bitcoin/

The total network hashrate in petaflop is 1719.  :o


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: RodeoX on June 13, 2013, 08:05:40 PM
No.


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: justusranvier on June 13, 2013, 08:13:04 PM
Maybe if it was a 50 exaflop supercomputer.


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: greyhawk on June 13, 2013, 08:21:16 PM
Wonder what's the power consumption of that beastie and if all that powers being used for ecologically sound purposes ;)

Like producing bitcoin?


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: threeip on June 13, 2013, 08:21:51 PM
How many 79xx does it have?  8)


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: BitCointransfers on June 13, 2013, 08:33:46 PM
Wonder what's the power consumption of that beastie and if all that powers being used for ecologically sound purposes ;)

The army is concerned with killing people more efficiently.


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: matthewh3 on June 13, 2013, 08:36:14 PM
Wonder what's the power consumption of that beastie and if all that powers being used for ecologically sound purposes ;)

The army is concerned with killing people more efficiently.

I think winning wars comes above that.  Example the game theory of mathematicians like John Nash during the cold war.


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: esenminer on June 13, 2013, 09:52:06 PM
If petaflops were the only thing determining hashing speed - which I don't think it is because SHA relies more on integer operations than floating point ones - looking at this

http://en.wikipedia.org/wiki/Radeon_HD_7000_Series

there is a column called GFLOPS - i guess gigaflops and the 7950 has 2867 gigaflops or .002867 petaflops.

so 50 petaflops = 17440 AMD 7950s = 17440 x 600 MH/s = 10.5 TerraHashes per second


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: threeip on June 13, 2013, 09:57:02 PM
If petaflops were the only thing determining hashing speed - which I don't think it is because SHA relies more on integer operations than floating point ones - looking at this

http://en.wikipedia.org/wiki/Radeon_HD_7000_Series

there is a column called GFLOPS - i guess gigaflops and the 7950 has 2867 gigaflops or .002867 petaflops.

so 50 petaflops = 17440 AMD 7950s = 17440 x 600 MH/s = 10.5 TerraHashes per second

I think this matters;
Quote
Usage of RISC SIMD instructions for GPGPU

Bloomberg etc use Tesla for number-crunching (http://techreport.com/news/17671/tesla-gpus-find-home-in-bloomberg-server-farm). I would be more worried about them.

Quote
Tesla K20X Peak single precision floating point performance 3.95 Tflops
(50 petaFLOPS)/(3.950 teraFLOPS) = 12,600 @ $13,000 = $163million

Quote
Assume BGQ rack is $2-3 M range and performance of 50 GFlops/node
http://www.usqcd.org/meetings/allHands2012/slides/Mawhinney.pdf (http://www.usqcd.org/meetings/allHands2012/slides/Mawhinney.pdf)

So the CUDA version is quite a bit more expensive, but probably works 163x better.


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: tinus42 on June 13, 2013, 10:54:10 PM
http://www.extremetech.com/extreme/155636-the-bitcoin-network-outperforms-the-top-500-supercomputers-combined


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: Littleshop on June 13, 2013, 11:45:58 PM
Petaflops are totally useless for mining  ;D

It might be better at scrypt mining.  It could even possibly do a 51% attack on all of the scrypt coin chains. 


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: jdbtracker on June 14, 2013, 12:22:48 AM
This is a data center for the NSA, they focus on encrypting and decrypting data... I bet you those data centers are full of Purpose built ASIC SHA2-3 AES etc encryption chips.


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: Elwar on June 14, 2013, 12:36:02 AM
The IBM technology is software based, not hardware based. It decentralizes the computations at a software program level so that you can run the program spread out over many machines. This works really well for feeding data through it in real time and not having any bottle necks. I used to work for IBM and took a class on how to use their Watson technology.

As for mining, saying that this is something to worry about would be like saying that we should worry about them using 100-1000 computers using CPUs to do a 51% attack.

But when I took the class they mentioned that they were able to feed all of the data of the Internet through the thing in 2 weeks. They tried to see how much data could break it in a simulation and their data producing engines could not create data fast enough to overwhelm it.


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: AbsoluteZero on June 14, 2013, 03:34:19 AM
Should Bitcoin fear the Army's new computer?

BITCOIN DOES NOT GIVE A SHIT.


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: Come-from-Beyond on June 14, 2013, 07:23:28 AM
Should Bitcoin fear the Army's new computer?

BITCOIN DOES NOT GIVE A SHIT.


Blissful ignorance...


Title: Re: Should Bitcoin fear Army's new 50 petaflops supercomputer?
Post by: freedomno1 on June 14, 2013, 07:26:00 AM
Petaflops are totally useless for mining  ;D

It might be better at scrypt mining.  It could even possibly do a 51% attack on all of the scrypt coin chains.  

Lol that

We mine more than the top 500 supercomputers on bitcoin could :)
http://qz.com/84056/the-bitcoin-network-is-now-more-powerful-than-the-top-500-supercomputers-combined/