Thx. So I take it that 24.6 is a later version of SOBPM?

24.6 subsumes the additional functionality of SOBPM1.tgz Look at the whatsnew.txt that comes with 24.6 for details.

I downloaded it and unzipped it and it only contained PRIME95.EXE, no whatsnew.txt or anything else. I am 98.75% finished P262144 stage 1 to 2,048,000,000. This is on a P4 3.4 Ghz. Could I switch to v24.6 and still use continue to a higher p-1? I think I could by just transferring in the save file (or is it files?). I think the save file is l0262144, am I correct? I hope I wouldn't have to start over if I wanted to use the new version. Would it be safer to just switch to p-1 on F25 or F26?

Maybe I should have posted this in that beta test thread? :no:

When you change over the .exe there won't be anymore need to start over than there would be if you decided to turn off the machine for a minute. There is no other files in the .zip because it is a beta test (one with a limited set of 'compentent' users testing it in the real world).

The mprime files do have the whatsnew.txt. And you do not need to worry about restarting. The savefile format has not changed!

I have done p-1 to b1=4B on F18! I sent the results to George, no new factors. This new version found all the known factors as well as the old one. It was much faster though on my P4. It wouldn't let me go higher though. Even when I tried to doo 4.1B it said no higher than 4B was allowed. Could someone walk me though continuing with gmp-ecm for p-1? I downloaded the pm1dump.2.zip. I do not have a compiler or the gmp-ecm program. Thx. P.S. I have never compiled before.

You could zip the Prime95 save file and send it to me (geoff AT hisplace DOT co DOT nz) and I will send you back the gmp-ecm resume file. You can get a gmp-ecm binary optimised for the P4 at [url]http://www.geocities.com/greatpsycho/[/url].

Did it. Did u receive it?

Joshua,

I wouldn't go beyond B1=4G with GMP-ECM, it is very much slower than Prime95 for base 2 numbers this large and the chance of finding a factor by increasing P-1 bounds that are already this large is very small anyways.

If you have a real lot of memory and want to do a stage 2, you might like to try the CVS version of GMP-ECM which has speedups in stage 2 for Fermat numbers. To get decent performance you should have at least 512MB available for the ecm process, preferably more. Btw, GMP-ECM 5.0 stage 2 uses as much memory, but is much slower for Fermat numbers.

Alex

I think I will only continue the stage 2 then. This is p-1 though not ecm. Where can i get the CVS version? I have the ecm5_p6_sse2 version now. I have 1GB of memory in my 3.4 P4. I also have an older P3 1Ghz with 2Gb.

2GB - now we're talking. This should let you use dF=4096 which will use ~2.3GB memory for the process but the swapping won't be bad. For stage 2 with GMP-ECM, the P3 with 2GB should be at at least as fast as the P4 with 1GB.

I see in an earlier posting that you use Prime95.exe, so I suppose you use Windows. I haven't built the current CVS version on Windows yet and trying it would take more time than I can easily afford at the moment, especially if George's DWT is to be compiled in (much faster) - do you have Linux installed on the Pentium 3?

Alex