View Single Post
Old 2007-08-03, 07:23   #52
Citrix's Avatar
Jun 2003

2×787 Posts

Originally Posted by geoff View Post
I think you have a much better understanding of the algorithms than I do, wouldn't it would make more sense for you to implement the main algorithm? I could then help speed up some of the lower level routines if necessary. If I were to try to implement SPH I would probably start the programming from scratch anyway, the srsieve code has become a bit complicated.
Sounds like a good idea to me. I basically would need to borrow the mulmod code from srsieve, the rest would be better to write from scratch.

Maybe you can create an .obj or .lib or .dll file that I can use with my program that has pre-compiled functions for modular multiplications. Basically you will be writing a fast 64 bit library. I think several projects can benefit from this.
(This way I won't have to deal with any low level stuff)

On second thought, there is no point re-writing the code your program uses. I will just re-write the initial Sieve of Eratosthenes and the corresponding SPH/ CRT code needed. I think we can keep the rest of the code the same, there is no need to write the rest from scratch.

Last fiddled with by Citrix on 2007-08-04 at 20:49
Citrix is offline   Reply With Quote