-   Sierpinski/Riesel Base 5 (
-   -   It'd probably be best if I just sieved. (

jasong 2005-03-13 23:09

It'd probably be best if I just sieved.
Guys, I have a 2200+ Sempron, and my thought is that it would probably be best if I only sieved.

It takes me 51 minutes to sieve a billion, so if anyone wants to "hire" me, or give advice on how long to sieve a number, please post.

Also, if possible, I'd like to be able to upload files for people to prp on Intels.

Can anyone help me with these things?

Mystwalker 2005-03-13 23:26

How much RAM do you have?
If it is enough, maybe we could exchange some work.
I have a Pentium4 for your PRPing, you could do some ECM work...

jasong 2005-03-14 00:26

Mystwalker, YGPM.

geoff 2005-03-14 03:40

Mystwalker beat me to it, I would love a Sempron to do ECM too :-)

But have you tried pfgw? I don't know for sure, but I think the very latest development version will be faster than PRP, its prp routines use the code from the latest Prime95 version 24.6 which had improvements for AMD chips.

(For SSE2 machines, PRP3 already has this code, and for P3 machines the new code might be slower which could make PRP the better choice, but I haven't tested this).

It is available in the files section of [url=]this[/url] yahoo group. The 2005 (alpha) version is the one with the 24.6 code.

As far as sieving for someone else with a SSE2 machine, I am not sure how to work out exactly how far to sieve on one machine, it is even more complicated when using different machines. But as a guide, using PRP3 a 2.66GHz P4 takes about 25 seconds to prp a number of the order 5^25000, about 6 minutes for 5^85000, and about one hour for 5^400000.

Maybe we need a benchmark thread.

All times are UTC. The time now is 15:25.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.