I notice that GIMPS exponents are tested with a bit of trial factoring and
then we go straight to LL testing, without any pseudoprime testing. I
assume that pseudoprime tests are actually slower than the LL test itself?
Or else they'd use them to eliminate some composite numbers before going on
to the LL tests...
(I am aware that Mersenne numbers are always strong 2-pseudoprimes. I was
thinking along the lines of other bases.)
--
.*. "Clouds are not spheres, mountains are not cones, coastlines are not
-() < circles, and bark is not smooth, nor does lightning travel in a
`*' straight line." -------------------------------------------------
-- B. Mandelbrot |http://surf.to/pgd.net
_____________________ ____|________ Paul Derbyshire [EMAIL PROTECTED]
Programmer & Humanist|ICQ: 10423848|
