Looks like things are resolved - I'm more than halfway through the 64K FFT's self-test, and no errors so far. Probably, again, the tiny exponents were sitting totally in the L2 cache (or being prefetched to it) and thus were putting more load on the CPU than normal LL work would do.
If M1257787 and M1398269 test out to be prime, I think I'll put my machine back in service (if only for trial-factoring, for a week or two). If they turn out to be composite, I'll repeat the process, and if either of them is still composite, well, I guess I'll be famous :) Nathan _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
