> I've heard several people question Davis's statistics from that > paper. However, regardless, the infrastructure I described before for > /dev/random can be applied to sample disk timings as well. How about a diskless system? -jm -- ---> http://www.jmp.fi/~jmarin/ <---