[EMAIL PROTECTED] wrote: > > On 26 Jan 2002, at 14:45, John R Pierce wrote: > > In any case it is rather doubtful that the compiler could make much > difference to Prime95. The pieces coded in C are of the "run once" > variety rather than being executed in every iteration.
That is true for LL testing and trial factoring, but not for ECM and P-1 on small exponents. Quite a bit of stuff is done in C there, notably generating primes <= B2. A better compiler might save a minutes or two per curve at B1=44M. But this is of course irrelevant to GIMPS as a record prime search, where the exponents are so large that even in P-1 at least 99.9% of the run time is spent in the asm-coded FFT. Alex _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
