[RC5] time spent on RC5-72

Marcin Sochacki wanted at gnu.univ.gda.pl
Tue Aug 28 15:07:29 EDT 2007


On Sun, Aug 26, 2007 at 08:25:16PM -0400, Anonymous wrote:
> the secret message was, as I'm sure a lot of others are too. Even
> without the prize money, the reason behind the contest still is valid:
> to find whether RC5-72 is strong enough encryption to survive today's
> computing power.

Let's assume 6 months as a reasonable time to crack a key and still be
useful. Most security practices recommend changing your
password/encryption key more often, but let's stay on the long end.

How many PS3s would it take to search the whole RC5-72 space?

My little calculation:
2^28 keys/s -- max rate of the PS3 (that's 268 Mkeys/s)
2^24 seconds -- slightly over 6 months

72-28-24=20

2^20 PS3s -- over 1 million

You'd need a cluster of 1 million PS3 (or equivalent) working non-stop
to crack RC5-72 in 6 months.

In other words, we've proven already that it's freaking hard to
brute-force RC5-72 with today's computing power in reasonable time. A
big company or a government can probably afford that. A volunteer project
surely not.

The real challenge, as I see it, is to write the client to achieve the
highest possible rate on a given CPU. Hats off to the d.net team for
hacking the assembly cores for so many architectures.

Once the keyrate of all the major CPUs is known, the actual cracking
becomes less important from the research POV. It's still fun for some,
of course :)

Wanted

-- 
+---------------------------------------+
|  -o)  http://wanted.eu.org/
|  /\\  Message void if penguin violated
+ _\_V  Don't mess with the penguin


More information about the rc5 mailing list