[Hardware] Other strategies for cracking RC5 ... [Was: Has anyone made progress on that CUDA client?]

John L. Bass jbass at dmsd.com
Sun Nov 23 13:53:43 EST 2008


Martin wrote:
>  I'm skeptical, but that's my nature
I'm curious why you would be skeptical.

If it's simply because "everybody knows" that RC5 is unbreakable, and 
because of that everybody stops trying to be creative and innovative in 
postulating potential new ways to attack the challenging problem, then 
yes, it's very unlikely it will be broken in the near term. That doesn't 
in anyway change how safe peoples data is that are using RC5 (or similar 
algorithms).

If it's because you have some experience that suggests that selected 
sboxes will not "leak" plain text data, or sbox configurations for plain 
text data, it might be worth sharing your reasoning.

Otherwise, the usefulness of d.net in attacking these problems, is that 
it sometimes takes a significant amount of data, and collaborative 
thought, to separate and correlate very small information leaks.

John


More information about the Hardware mailing list