On 1 Dec 2001, at 17:39, George Woltman wrote:

> This is because my rather limited reporting software only adds up the
> LL results in the verified and one-LL-tests databases.  Once an
> exponent is factored it is removed from those databases.

The other problem here is that the "known factors" database does 
not include the discoverer.
> 
> I prefer a factor to a double-check.  But it is hard to quantify
> "prefer" in a mathematical formula for computing trial factoring
> limits.  Prime95 uses the formula:   cost_of_factoring must be less
> than chance_of_finding_a_factor times 2.03 * the cost_of_an_LL_test.
> 
> This should maximize GIMPS throughput.  The 2.03 is because we must
> run two (or more) LL tests to do a double-check.

Again there is a complication, since the ratio of time to do factoring 
assignment X to time to do LL/DC assignment Y varies according 
to the processor. e.g. PIIs are relatively quick at factoring, whereas 
P4s are much more efficient LL testers.
> 
> P.S.  I'll comment on the M#39 news later.  For now lets celebrate our
> grand accomplishment rather than worry about non-optimal press
> coverage.

Hear hear. Congratulations to the discoverer (whoever he/she is), to 
George on his program finding a fifth Mersenne Prime, and to 
everyone involved in the project, without whom we wouldn't have 
reached this milestone.


Regards
Brian Beesley
_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ      -- http://www.tasam.com/~lrwiman/FAQ-mers

Reply via email to