-   Software (
-   -   How to only do Trial Factoring? (

michael 2003-12-19 23:48

How to only do Trial Factoring?
Do i have to adjust settings somewhere if i just want to do trialfactoring? Or should i abbandon when it starts with an LL-test? Also could i adjust the bits uptil where it factors? When i check the status, i see some numbers have been checked for factors up to 72 bits, what is optimal for checking? And how long would it take on a pentium 4 2.4Ghz to trial factor up to 72 bits (regarding you don't find a factor)


outlnder 2003-12-19 23:54

The top side factoring bits are built into the program, so you don't need to change them. If you want to, look into undoc.txt in your prime95 folder. But it is best to leave them as is.

To select trial factoring only, open the prime95 window, go to "Test", click on "PrimeNet" and then check ONLY the "Request Mersenne numbers to Factor".

This sets you up to do trial factoring only.

GP2 2003-12-20 01:36

[QUOTE][i]Originally posted by outlnder [/i]
[B]This sets you up to do trial factoring only. [/B][/QUOTE]

However, that only applies to future work.

If your computer is currently doing an LL test, it will continue that to completion unless you unreserve it (using the advanced features of the Prime95 menu).

michael 2003-12-20 01:47

I don't have advanced options ... yet? How do i get them?


outlnder 2003-12-20 05:17

Go to "Advanced" and you will see "Password", type in 9876 and hit enter. Then go back in and you will see the "Advanced" menu.

Just for fun, how much work have you completed on the LL test?

michael 2003-12-20 08:42

Done around 4 million iterations, I will complete this test...
I just thought i'd be more productive doing trialfactoring, i guess i need to play a bit with the program, look around for the options and see what suits me best. Just that now it started with the Lucas-Lehmer test it looks like i'm hooked to that for now, can't do anything else...
Also i'm not too sure of the reliability of my hardware. First time i did the torture test it came up with an error, my computer had been running for a few days straight so i rebooted and after that it went through the test smoothly...still i'm not too confident. If i just stick to trialfactoring it's easy to check if sthing goes wrong, to see if something went wrong during an LL test, i can only think of redoing the LL-test completely to figure that one out.

markr 2003-12-20 10:22

If you're at all doubtful about your hardware, you could do some double-checks. That will give a very good confirmation of how reliably your computer completes LL tests. Your choices for the current assignment would then be: complete it before doing double-checks, do double-checks and later on pick it up where you left off, or abandon it (which seems a waste to me, unless there were errors).

Alternatively, you could get some reassurance by running prime95's torture test for a day, or use memtest86 to check your memory.

Pentium 4s are great for LL tests - first-time or double-checks! (Of course they're also very good at trial factoring above 64 bits, which is where most of the time will be spent anyway.)

Most current trial-factoring assignments will be done to 67 bits. I would guess a 2.4GHz P4 would complete one in 1.5 days. Of course not actually having one myself :sad: someone else will probably give a more accurate estimate.

Sorry if there are too many choices here! See what suits you best, as you say.


PrimeCruncher 2003-12-20 15:11

[QUOTE][i]Originally posted by markr [/i]
[B]Most current trial-factoring assignments will be done to 67 bits. I would guess a 2.4GHz P4 would complete one in 1.5 days. Of course not actually having one myself :sad: someone else will probably give a more accurate estimate.[/B][/QUOTE]

Given that TF is a function of clock speed, I have estimated, based on my Celeron 2.0GHz (which takes ~32 hours to complete a TF), that a 2.4GHz P4 should take ~25 hours to complete a 2^67 TF with no factor.

Unregistered 2003-12-21 06:01

The interesting part is if you trial factor to 70 bits instead of 67. The time involved is about quadrupled.

One of these days I need to take a close look at George's algorithm to eliminate potential factors. He eliminates roughly 19 out of 20 possible factors by 2 simple tests. That still leaves *illions of potential factors (* depends on how big the exponent is)


michael 2003-12-21 11:39

If a prime q is a divisor of Mp it must be q=2kp + 1 and q=+/-1mod8

I don't know of any more eliminations for factors, and i'm pretty sure this is built into the program here.


QuintLeo 2003-12-25 03:54

Above 64bits, each extra "bit" of trial factoring takes twice as long to complete as the bit before it. Up to 64 varies somewhat, due to usage and availability of registers on various CPU types, and other lesser factors....

All times are UTC. The time now is 23:53.

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