[RC5] Interesting Project... MD5

Malte S. Stretz msquadrat.nospamplease at gmx.net
Fri Sep 27 18:25:02 EDT 2002


On Friday 27 September 2002 16:40 CET Francis Julien wrote:
> I would like to know if this kind of project is possible with
> distributed.net?
>
>[...]
>
> More info here -> http://www.certainkey.com/dnet/

>From JL's paper:
| 2. HOW LONG WILL THIS TAKE?
| ---------------------------------------
| Our prototype client gets about 5 MegaMD5s/sec on an AthlonXP 1700,
| comparable to the same number of RC5-64 MegaKeys/sec on the same
| processor.
| 
| 264 MD5 hash's will need to be calculated before we can expect to have a
| collision. This is due to the birthday paradox.  With 127.24 GigaKeys/sec
| == 127.24 GigaMD5s/sec the same speed as the RC5-64 project, and
|  extrapolating Moore's law, 264 MD5s will be reached in 2 years.

But who's running dnetc on an Athlon 1700? (Is he talking aboout 1700 MHz or 
1700+ btw?) Most of us got it IMO running on their servers or other oldish 
boxen which _don't_ follow Moore's law. So leaving good ol Moore out of 
this calculation, I get:

2^64 keys / 102437151000 keys/s ~= 180079000 s
180079000 s / 3600 s            ~= 50022 h
50022h / 24 h                   ~= 2084 d
2084 d / 356 d                  ~= 5.85 y

Let's assume Moore is also compensated by the contributor fluctuation 
lately. Then we'll have our first collision in approx. 5 years, not in 2.

Don't understand me wrong: I kinda like that project. But I like to be 
realistic, too ;-)

Malte
-- 
--- Coding is art.
-- 


--
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