Previous posters wrote about how some exponents jumped around between 1
and 15 a lot, while they "should" progress steadily forward, reaching
level 9 as they start factoring the final bit.
George asked if perhaps this was a bug in an old version.
No: each version of the software I have used for factoring (17, 18, 19,
and 21) worked this way. BUT:
Only a computer that is on and connected more or less continuously to the
internet will report reliably. If a computer is past due for a report when
it is booted, it always reports 1 for factoring progress regardless of
actual progress made. ("Connecting primenet server" happens *before* work
resumes on the exponent itself, and the program doesn't bother to check
where it is, I think. Curiously, LL tests don't suffer the same glitch.)
This was a great irritation to me a couple years ago, on a Cyrix that was
used erratically (sometimes off for a week, sometimes used a few hours
each day, sometimes left on overnight): I was unable to keep a 2nd
exponent queued reliably, because the rolling average was so low that it
would estimate months to complete an exponent, see that "1", and unreserve
any work that was waiting, even if it was in fact going to finish
the current exponent the same day.
GRB
_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
