Jabari Zakiya wrote: > If someone has a faster coded > implementation the SoA > I would appreciate seeing it, so I can benchmark it against ALL my > versions of my SoZ.
http://cr.yp.to/primegen.html Regards, Antonio ------------------------ Antonio, Thanks for the link to the code. I actually found this code from my initial searching, but I didn't know it was for the SoA. Since I'm not into C, I couldn't follow it by just looking at it, and haven't tried to decipher it yet. Are there any C/C++ experts willing to translate my SoZ versions and run it against this C SoA code? It should be A LOT easier to do than the SoA, and much shorter code. This is just a quick note FYI. When I timed M9 = (2**61)-1, (M9.primz?) with my original primz? it took about 92 minutes, on my P4 2.8 GHz laptop. Then I coded it using the 16 P60 residues, and it got faster, and then I did P7. Then I had one of those "duh" moments and realized I was doing a lot of excess work, and made it even faster. Sleep and concentration are real helps. This new version reduced the time from 92 minutes to 7+ minutes (~448 seconds) for M9. These are all method/algorithm enhancements, which I'm continuing to investigate. I haven't even begun to think about coding enhancements. However, some of the numerical methods optimizations used to do modulo math, et al, are also applicable to implementation enhancements to my algorithm. I hope others will begin to apply this wealth of knowledge and know-how to my method. Jabari -- See Exclusive Videos: 10th Annual Young Hollywood Awards http://www.hollywoodlife.net/younghollywoodawards2008/ _______________________________________________ Prime mailing list [email protected] http://hogranch.com/mailman/listinfo/prime
