Brian J. Beesley wrote: > Shouldn't make much difference Of course.
By Daran's example (quoted here in reverse order), 7714127 B1=40000, B2=650000 7786567 B1=40000, B2=640000 one can easily see that the effect of differing FFT sizes (384K for exponent 7714127, 448K for 7786567) on choice of P-1 limits was small in this case. But Daran's question was about why the the _smaller_ exponent should get the _higher_ B2 limit at all. My point was that that was a result of the FFT difference on the algorithm's output. > P-1 computation rate is directly linked to LL testing computation > rate, since the same crossover points are used and the main cost > of both is FFT multiplication. True, but the question here was about the choice of B1 and B2 limits. > I think the most likely cause is that the P-1 limits depend on the > relative speed of P-1/LL computation speed and trial factoring speed. No. Neither trial factoring speed nor its ratio to P-1/LL computation speed enters into the P-1 limit-choosing algorithm. > If RollingAverage was slightly different, the relative computation > rate might be "guessed" to be different enough for the P-1 limits > to vary slightly. No. The P-1 limit-choosing algorithm does not consider RollingAverage either. Richard Woods P.S. When replying to this message, check that your reply's To: field contains "[EMAIL PROTECTED]". My ISP's mail handler has been mangling the Reply-To: field of messages I send. _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
