[rc5] How long this could take....

Rik Ling rling at pipcom.com
Sun Aug 17 22:33:59 EDT 1997


Two items of note.

1)  The banner ad on Altavista actually works, 'cause it sucked me in while
I was in the middle of a search for some CD-ROM drivers.  If not for it I
never would have known about the RC5 contest and certainly not that
distributed.net even existed.  I was wondering how many other participants
joined as the result of seeing an online ad?  Good idea, whoever's it was. 
I run an ISP with six Pentium and five 486 machines on premises, plus my
Pentium at home--all told an addition of about a million keys per second of
processing power that is now devoted to cracking RC5.  Not too bad a return
for one lousy banner ad, eh?

2) I'm new to the cracking effort, and have been running some of the
numbers through my head.  Correct me if I'm wrong.

Rate of key examination on Aug 16:	3,025,448,000 keys/sec
Percentage of keyspace exhausted:	14.4%

(Taken from stats page as of approx 23:00 GMT, Aug 17th, 1997)

According to my trusty calculator, this means we have
61,681,300,000,000,000 (quadrillion?) keys left to search, give or take a
few billion.  At a constant rate of about 3 billion per second, it will
take another 20,387,493 seconds to examine the remaining keyspace.  That's
235 days of bit-bashing if we're unlucky and the magic key is somewhere
near the end of the keyspace, or (heaven forbid) was missed due to an
unreturned keyblock.

Now, for what I see as good news:

1) If we assume the key will be found at the 50% mark in the keyspace, we
only have another 98 days to go.  Not very long for me to get my team into
the top 100! :-)

2) All these calculations assume a constant processing rate, which is
obviously not the case.  The rate of processing has been increasing (basic
assumption on my part) and will continue to do so.  Has anyone tried to
calculate a rate of acceleration for the keys/sec? If they have, it would
be interesting to throw it into the mix and see what happens to the
projections.  Is the Bovine RC5 project experiencing exponential growth? 
Linear growth?  Logarithmic?

The project has been active for 146 days now, so if we assume constant
growth per day (not true, but live with it for now), then the keys/sec rate
has been increasing by 20,722,000 keys/sec/day -- this means we will be
examining 1,790,380,800,000 more keys every day than we did the previous
day. Eeep! We have already examined 10,358,000,000,000,000 keys, and we are
currently examining them at a rate of 261,398,700,000,000 per day
(3,025,448,000 keys/sec * 60 seconds * 60 minutes * 24 hours).

Hmmm, let me grab my spreadsheet....plug in the numbers....project
ahead....

According to my very approximate math, we'll hit the halfway point in
another 80 days, rather than my original 98, and we'll exhaust the entire
keyspace in only 157 more days, instead of 235. We could have this thing in
the bag before Xmas. (Doncha just love cumulative growth?)

I hope I haven't just re-hashed something that has already been discussed
to death.  If so, I implore you to ignore this post and just send me a
teeny message to the tune of "been there, sonny".  If this isn't old news,
then I'd really like to hear from someone who can calculate an approximate
growth curve for the projects overall processing power. 

---
Rik Ling 
It's a mad mad world.  Only the fully sane are truly crazy.
rling at pipcom.com

----
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.



More information about the rc5 mailing list