mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   PrimeNet (https://www.mersenneforum.org/forumdisplay.php?f=11)
-   -   How to manually reserve TF (https://www.mersenneforum.org/showthread.php?t=25211)

JuanTutors 2020-02-10 19:51

How to manually reserve TF
 
First, I know, GPU and all. I have an old 32 bit Lenovo laptop that I'm pretty sure can't do mfacto and mfactc. This laptop will be on MAYBE 50 hours per year. So I want to set it to do TF on a Mersenne I've already reserved. But I don't know how to do that. I see the worktodo does Factor=[HEX],exponent,lowerbound,upperbound . I don't know how to deal with that hex.

I asked a similar question about a Chromebook but I'm probably not going to do anything there.

chalsall 2020-02-10 19:58

[QUOTE=JuanTutors;537250]I don't know how to deal with that hex.[/QUOTE]

Change it to be "N/A", and then Prime95/mprime won't try to reserve it from Primenet, but will report the results once completed.

You're aware that TF'ing on a CPU is a waste of electrons, but they're yours to do with as you wish... :smile:

JuanTutors 2020-02-10 20:03

[QUOTE=chalsall;537252]Change it to be "N/A", and then Prime95/mprime won't try to reserve it from Primenet, but will report the results once completed.

You're aware that TF'ing on a CPU is a waste of electrons, but they're yours to do with as you wish... :smile:[/QUOTE]

It's that or nada for a 7 year old laptop that will be on for minutes per day.

JuanTutors 2020-02-10 20:14

I know it's greedy to use the tiny number of minutes I am running this to work on my own Mersennes but honestly when I say max 50 hours, it's more like "I haven't turned on this laptop in 2 years" minutes. It's just going to do occasional calculations, and periodic P-1 on my Mersennes will require time to do GCD, and the chance of the laptop being on that long is low. Unless I'm wrong about this laptop being able to do mfacto/mfactc and unless that would be faster on this old laptop, that seems to be my option.

petrw1 2020-02-11 00:31

[QUOTE=JuanTutors;537255]I know it's greedy to use the tiny number of minutes I am running this to work on my own Mersennes but honestly when I say max 50 hours, it's more like "I haven't turned on this laptop in 2 years" minutes. It's just going to do occasional calculations, and periodic P-1 on my Mersennes will require time to do GCD, and the chance of the laptop being on that long is low. Unless I'm wrong about this laptop being able to do mfacto/mfactc and unless that would be faster on this old laptop, that seems to be my option.[/QUOTE]

For low hours PCs P-1 or ECM are your best bet. ECM assignments will finish in hours , P-1 in a couple days.

Uncwilly 2020-02-11 00:42

PRP on cofactors is a fast operation too.

JuanTutors 2020-02-11 02:34

[QUOTE=petrw1;537271]For low hours PCs P-1 or ECM are your best bet. ECM assignments will finish in hours , P-1 in a couple days.[/QUOTE]

So since I've already done P-1 on a higher ram and faster computer and have done TF to 2^81, I should do ECM first and then TF?

VBCurtis 2020-02-11 05:18

You shouldn't do TF, period. ECM is fine, or just accept that that particular machine isn't a useful contributor to GIMPS- just like the machine you carry in your pocket daily.

snme2pm1 2020-02-12 11:35

Indeed 50 hours per year is of little benefit to the project, and any assignment that you may have held from the outset would likely expire before completion, and likely be completed by another player.

JuanTutors 2020-02-12 12:31

[QUOTE=snme2pm1;537416]Indeed 50 hours per year is of little benefit to the project, and any assignment that you may have held from the outset would likely expire before completion, and likely be completed by another player.[/QUOTE]
First, really posting to find out if the error I got above is a bug in Prime95, since it worked fine on a smaller exponent.

Second, again, just doing some light TF and ECM on a project I'm already doing PRP with on another computer on the off chance I find a factor and save me whatever time remains from the PRP test. From what I gather, the probability of finding a factor between 2^81 and 2^82 is slightly above 1%, which is greater than 0%. I get that it's slow, but I haven't heard an actual reason why I should opt for no work over some work, when any work is positive work.

Uncwilly 2020-02-12 14:02

[QUOTE=JuanTutors;537418]First, really posting to find out if the error I got above is a bug in Prime95, since it worked fine on a smaller exponent. .[/QUOTE] You posted that in a different thread.

But, what people are trying to tell you is that some work is really of zero value. Doing ECM factoring on a number that is years ahead of the first time wave front is nonsensical. ECM is best used for [B]low numbers[/B] that have had a good amount of TF and P-1 done.

If you want to make maximum use of your machine for the GIMPS project overall, I would suggest that you set it to do PRP (or PRP-DC) on Mersenne number cofactors.
The reasons are:
It is not a huge amount of work in terms of GHz days per assignment.
You will not be holding up any milestones.
Your assignment will not likely expire after a couple of months of slow progress.
It will help the project along, by revealing an important bit of data.
If your test does not match, a TC can determine if your machine is still sane.

Just because you are obsessed with a different number, doing any testing you can on it does not make sense. The only reason to do extra TF or P-1 on a number that is undergoing PRP or LL testing, is to try to "break" an number that is claimed to be prime, while the DC is being done. Doing ECM on such a number is fool-hearty. It is like trying to empty a supertanker with a bucket.


All times are UTC. The time now is 23:32.

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