Re: Mersenne: Question about speed...
Hi, At 12:42 AM 6/21/99 +0200, Otto Bruggeman wrote: >Sorry to bother you people with this but can anybody tell me why my celeron >400 all of a sudden slows to (almost) half speed when it reaches the >1,000,000,000 (actually a little more, i guess it's around 2^30) mark in >factoring numbers??? My p233mmx slows about 15 percent. The numbers >currently being factored are in the 9.7 million range (shoudn't matter, it >happened also in the 7.5 million range...). This is normal. The factoring code uses different algorithms to find factors below 2^62, from 2^62 to 2^63, and from 2^63 to 2^64. What you have noticed is prime95 switching over to the 2^62 to 2^63 code which is slower than the below 2^62 code. Hope that helps, George Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Re: Mersenne: Question about speed...
> >Sorry to bother you people with this but can anybody tell me why my celeron >400 all of a sudden slows to (almost) half speed when it reaches the >1,000,000,000 (actually a little more, i guess it's around 2^30) mark in >factoring numbers??? My p233mmx slows about 15 percent. The numbers >currently being factored are in the 9.7 million range (shoudn't matter, it >happened also in the 7.5 million range...). > >Thanks in advance... > >Otto. > Prime95 is trying to find factor in the 2^63 range, soo it take a bit longer then 2^62 1,000,000,000 * 2^32 ~= 2^62 David __ Get Your Private, Free Email at http://www.hotmail.com Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Re: Mersenne: Question about speed...
At 08:41 PM 6/20/99 -0400, lrwiman wrote: > >Doubtful, since the LL remainder never goes above Mp. The only thing that >inherantly increases is the iteration count, which takes up about log_2(p) >bits. Not that much... >P.S. 2^30 is ~100,000,000 >not ~1,000,000 Actually 2^30 is ~ 1,000,000,000. 2^20 is ~ 1,000,000. He said 1,000,000,000 but he may have meant 1,000,000. Prime95 can't handle 1,000,000,000 yet. Anyhow, the Celeron cache is only 128KB, which is 1.34 million bits. So the Celeron cache size may be the reason for the slowdown at about 1,000,000 bits. +--+ | Jud "program first and think later" McCranie | +--+ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Re: Mersenne: Question about speed...
>> Sorry to bother you people with this but can anybody tell me why my celeron >> 400 all of a sudden slows to (almost) half speed when it reaches the >> 1,000,000,000 (actually a little more, i guess it's around 2^30) mark in >> factoring numbers??? My p233mmx slows about 15 percent. > I'm just guessing, but it could be that the Celeron's smaller cache gets > filled up. Doubtful, since the LL remainder never goes above Mp. The only thing that inherantly increases is the iteration count, which takes up about log_2(p) bits. Not that much... -Lucas Wiman P.S. 2^30 is ~100,000,000 not ~1,000,000 Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Re: Mersenne: Question about speed...
At 12:42 AM 6/21/99 +0200, Otto Bruggeman wrote: > >Sorry to bother you people with this but can anybody tell me why my celeron >400 all of a sudden slows to (almost) half speed when it reaches the >1,000,000,000 (actually a little more, i guess it's around 2^30) mark in >factoring numbers??? My p233mmx slows about 15 percent. I'm just guessing, but it could be that the Celeron's smaller cache gets filled up. +--+ | Jud "program first and think later" McCranie | +--+ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
