On Tue, Dec 30, 2008 at 4:27 PM, <[email protected]> wrote: > On Tue, 30 Dec 2008 16:19:31 PST, chort said: > >> The viability of that approach depends on how much the code depends on >> the systems being clustered together over low-latency interconnects. >> 4000 machines spread all across the internet separated by 300ms of >> latency is not the same thing as 4000 machines in the same room >> running a cluster OS. > > They give a hint that it's *highly* parallel code: > > "This part is not suited for the PS3s SPU cores due to the large memory > demands > and the high number of branches in the software execution flow." > > Presumably, if the hit that a lot of branches create is bad, the *huge* > hit of even an Infiniband interconnect would be fatal... >
Ah, you're nit-picking on the fact that I mistakenly mentioned the actual collision rather than the birthday attack, so yes my bad for being careless with my terminology. The more time-consuming part of the computation was the birthday attack, which is what the PS3s _are_ good at. You're right that the collision blocks worked better on machines with more RAM and instruction sets/pipelines designed for more branching. In any case, we don't disagree that it's possible to conduct the attack with a moderate-sized botnet. We also agree that nutd0rk has no idea what he's talking about (not that he ever does), so this discussion seems to be heading no where. -- chort _______________________________________________ Full-Disclosure - We believe in it. Charter: http://lists.grok.org.uk/full-disclosure-charter.html Hosted and sponsored by Secunia - http://secunia.com/
