> > Well, the current range of Mp candidate values (just under 8,000,000) is
> > described as using a 63 bit 'factor'... I would assume eventually this
has
> > to go to 64 bits, but I'm not clear on the relationship between the
exponent
> > and the FFT size and the 'fact bits' listed on the test status pages.
>
> OK, so you mean the exponent at which changeover from 63 bit trial
> factoring to 64 bit trial factoring occurs. This is a bit over 9
> million. The "break" actually occurs where we want to do more than 64
> bits trial factoring... George's current code is just as good at
> trial factoring to 64 bits as it is to 63 bits.
and the FFT sizes? If I recall, right now we are doing ... ah, I found the
table on George's site, duh.
at 7,820,000 we just crossed over into 448K FFTs.
at 9,090,000 we hit 512K
at 10,380,000 we hit 640k.
etc til > 20,500,000 exceeds 1024k
so how big of a FFT can the current 18.x.x Prime95 handle? Up to 1024K
FFTs? If so, I guess we still have some headroom...
-jrp
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
