All,

Sorry if my opening was confusing, I was trying to show where the thought
came from, not imply the methods were similar enough to intermix them. And
that led to the FFT uncertainty - ok, yes, I should have remembered that FFT
is better suited for 2^(2^64), not 2^64.  8*(

I see now how the number of inner products would grow rapidly. I overlooked
this because I wrote my test programs in Pascal, using a numeric type called
comp, which is a strange type of 4 word integer. So both my squaring and
modulus subroutines used a two-step inner product - I just had to do the
doubling separately to avoid getting an overflow.

I am glad that my question sparked some thought. When I first read the math
sections, I was surprised at how small the sieve was for Prime95. I just
figured it was shared between the trial factoring, the P-1, and the LL code
- making memory the first consideration. I guess access time would be the
second limitation.

Out of curiosity, what are the memory expenses of the trial factoring code?
Does it store a sieve other than the 120 block in the description? Does it
store a sieved list of potential factors? Or does it determine the next
factor as it goes along? The description in the faq kind of sounds like a
little of each. If there's already somewhere to find this info, please tell
me where to go. ;^}

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

Reply via email to