mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Information & Answers (https://www.mersenneforum.org/forumdisplay.php?f=38)
-   -   Stat Summary (https://www.mersenneforum.org/showthread.php?t=15640)

Unregistered 2011-06-04 21:16

Stat Summary
 
I've been running GIMPS for about three hours now but on the "My account" page on the website no stats appear. I don't really care if the stats ever show up, but I can't even tell if the program is running.

Uncwilly 2011-06-05 05:28

1 Attachment(s)
[QUOTE=Unregistered;263005]I've been running GIMPS for about three hours now but on the "My account" page on the website no stats appear. I don't really care if the stats ever show up, but I can't even tell if the program is running.[/QUOTE]Your stats will change when your machine turns in a completed assignment. That is usually happens between 1 and 30 days (depends on the machine and assignment type.)

Do you see a green icon like the one below in your systray? If you do Prime95 is running. If you don't see it, it still might be running. Check in TaskManager for the Prime95 process. If you see it, it is running.

Christenson 2011-06-05 14:05

On non-windows boxes, run ps -aux to check for the prime95 process.

On windows boxes, which I ASSUME we are talking about here, double-cllick on the Prime95 icon to bring up a window with periodic progress updates. I like to chop a zero off of the "iterations between status updates" field in the menus.

If you want stats quickly, do P-1; which should get you stats feedback in 3 days or so; if you are really impatient, do a little ECM on small numbers and give it about 50M of memory for ECM; this will complete a task in a few hours.

P-1 is the most effective activity from the standpoint of making progress on the project; I'm running about 60GHz days per factor found; each factor found eliminates 2 LL tests at 80GHz days apiece. However, only first-time LL tests will actually find Mersenne prime #48, barring a really lucky roll of the dice and having an LL-D find a prime after someone else's incorrect first-time LL test.

Mr. P-1 2011-06-07 17:34

[QUOTE=Christenson;263048]If you want stats quickly, do P-1[/QUOTE]

So long as you allow it sufficient memory. TF also allows quick entry onto the stats chart and doesn't require much memory.

[QUOTE]P-1 is the most effective activity from the standpoint of making progress on the project; I'm running about 60GHz days per factor found; each factor found eliminates 2 LL tests at 80GHz days apiece.[/QUOTE]

Provided you allow it sufficient memory, P-1 is indeed the most effective activity, but not for the reason you give. TF eliminates far more LL GHz days per TF GHz day expended than P-1.

But with TF, if you don't do it, someone else will. The factors you find would have been found anyway, so you're not really saving any LL tests. Your contribution is worth just the GHz days you expend.

With P-1, if you don't do it, someone else will, but [i]probably not as well[/i]. About half of all P-1s are not given a stage 2 at all. Many of those that are, are done to minimal bounds. By doing P-1 with adequate memory, you will find factors which otherwise would never have been found. Your contribution is worth the GHz Days the other guy would have expended [i]plus[/i] the LLs saved by the additional factors you find. That's worth a lot more than just the GHz days you expend.


All times are UTC. The time now is 12:26.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.