[RC5] I think we found our next contest...

Andreas D. Landmark andreas.landmark at noxtension.com
Fri Dec 14 00:09:15 EST 2001


At 13.12.2001 20:07, Aaron W. Swenson wrote:
>There are some things that aren't time sensitive (i.e.: blueprints, long 
>term plans, relations).  Also, if they have about $50,000, then they can 
>build a 64 node Linux cluster utilizing AMD Athlon processors.  That's 
>just as powerful as some $21 million super computers.  So, you don't need 
>a "**** load" of money to get a lot of processing power.   Okay, okay, it 
>seems like a lot to us, but not to some one/thing that has millions or 
>billions.

Well, you've missed an instrumental point about brute-force cracking of 
encryption.
The reason a $50,000 cluster *MIGHT* perform as good as some $21M supercomputer
is that for certain types of computations big computers doesn't scale as 
well as
a distributed effort.

A 64 Node cluster (given 2 cpus pr cluster) would running the dnetc client 
probably
perform pretty equal with a 128 CPU computer as all the CPUs are cracking at
full speed, there isn't any improvement in speed on bruteforce cracking by
distributing/centralizing the effort.

The only real difference is the practical matters such as maintaining the 
installation.


>And yeah, AES might take to long, but it won't be long before we can crack 
>that code in 22 hours as well (under ten years.)  I haven't seen anybody 
>mention the 64bit processor development.  Intel already has 64bit 
>processors, and AMD is developing theirs and should have some available by 
>2003.  Of course Intel's 64bit processors are aimed at the server market, 
>however, AMD's is aimed at the home market.

We've worked on rc5-64 for 4 years, and 4 years ago it wasn't near as 
uncompletable
as AES seems today. I think you'll have to, at least with conventional 
computing (keeping
quantum-computing et al out of this), double or almost triple your guestimate.

I don't know what numbers or nuggets of information you base your 
guestimate on, but
it seems to be taken out of nowhere and isn't supported by any information 
I've
ever seen...

The speed of bruteforcing cryptography is neither a priority when 
developing CPUs or
directly proportional to the development of new CPUs.

Take the PIV (and P3) for example which all are an improvement (in some 
respect) on
the preceding CPUs, but their improvement in rc5 rates where rater meager...

>I've herd the Dnet staff make mention about 64bit encryption fitting 
>nicely into two 32bit registers.  Correct me if I'm wrong, AES is a 128bit 
>encryption process, so it should fit nicely into two 64bit registers.

How many 64bit systems have you got access to?
         It would be many years until a "64 bit on every desk in every home".

There isn't a competition for AES announced as far as I know, and the 
scientific
return of cracking AES is rather sparse, it would prove that AES takes ages to
bruteforce even on a huge supercomputer such as dnet...

Results of bruteforcing has been proven before (and is currently being 
proven with
rc5-64).

When the rc5-64 project is finished, I wouldn't be a stranger to 
introducing projects
similar to those of UnitedDevices, cancer-research or other medicalbased 
research
is a good thing to support (even though I've got somewhat controversial 
opinions on
the evolution of modern medicine)...

-- 
Andreas D Landmark / noXtension
Real Time, adj.:
         Here and now, as opposed to fake time, which only occurs there
and then.

--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list