>I have one P166 noMMX. I did my first LL test (M5196497) at approx .57.
When
>this exponent was finished and prime95 loaded the second (M5307469),
the
>time/iter rose to .66 or so. Never been able to clear this up nor fix
it.
>Upgrading my win95 to 98 didn't make it run any slower or faster
though.
>Weird.
It's not a bug. 5.26 million is the upper limit of one of the FFT sizes;
the two exponents that you mention are on opposite sides of that
boundary. The larger one takes longer per iteration because the program
has to manipulate a 320K FFT instead of 256K.
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com
