GIMPS/primeNet CPU overview refresh rate

[Meral Harbes]

Hello GIMPS participants


I've started putting my systems into primeNet a few days ago. I'm not sure about my motivations as of now - it's probably a mix of a "thank you" for all the times I used prime95 as simple stress-testing software (I tragically didn't know its actual purpose before) and reducing heating costs to absolutely zero, since my holy chamber is currently manned by two calculators and another two Tetris accelerators which are all trying to make Vulcan proud, battling each other for his praise.

I have also enslaved my small but mighty and portable companion to enrage fellow students with it's soft, never-ending song of - as he calls it himself - "his people". Besides that primeNet currently thinks that this pocket-calculator is my most efficient calculator of all, which I guess it takes from it sending the most reports and having a greater total of GHz-hours than the other systems, since their tasks are not complete yet. What I actually want to inquire about is that primeNet has currently set it's "days to-go" to -2, which I assume is code for "hamburger", but that is not what I think I have ordered at all. Even though I have tried to implore it to set the expectation date further into the future, it ignores all my companion's requests and stays on hamburger, as if it's already being grilled by somebody else. Or is my interpretation of -2 wrong altogether? I'm just worried that somebody else is cooking the same meat right now. What a waste that would be.

Also I have allowed prime95 to pretty much use all the resources it wants, which it does with joy. But what does it actually put into 7.2GB of random access marmalade that it currently folds his greedy hands over? I just hope it doesn't spit in it, or mark it in another way to be his forever.


Prime regards

Meral Harbes

[lycorn]

Hi,

You see "-2" as the number of days to go because the test was expected to have fnished 2 days ago. When you reserve an exponent, Primenet computes the expected time to complete based on the exponent/assignment type, your hardware configuration and the number of hours per day Prime95 is supposed to run. If you switch off your computer, or stop Prime95 for a longer time than the one initially declared, your assignment will be "late" from Primenet´s point of view. When you communicate with Primenet, automatically or manually, it will recompute the time based on the current progress.
As for the resources made available to Prime95, don´t worry, at least while performing TF or LL/DC tests. The amount of memory taken is minimal. When doing P-1 or ECM tests, Prime95 will use significant amounts of memory during stage 2, so be careful not to allocate too much memory (on a 8GB system, I wouldn´t allocate more than 6GB).
Welcome to the gang and enjoy!

[NBtarheel_33]

Wow, we need to introduce this poster to Kathegetes! A match made in Heaven!

[Prime95]

Originally Posted by NBtarheel_33

Wow, we need to introduce this poster to Kathegetes! A match made in Heaven!

I found the post quite entertaining. A welcome addition to the forum.

Something may be amiss. Prime95's default configuration is to send expected completion dates every day. If this happens, one should never see an expected completion date of -2.

Assuming the machine has been on the last few days, a look at prime.log may indicate some kind of communication problem.

[TheMawn]

Regarding the random access marmalade, it will only use 7.2 Gravy Boats worth if you're running P-1 factoring or ECM factoring.

For instance, I'm using about an eighth of a Gravy Boat of marmalade across four workers.

[Meral Harbes]

No worries, lycorn. I won't overdo my mighty student blaster. He was indeed playing with P-1 factoring. I'm going to blink trough this weird library to find some info on as to what exactly it stores in that marmalade. For now I have, in blind believe that the prime god puts it to good use, prepared even more Gravy Boats to see if I may stop its hunger. While I had lost overview of which jobs were done and already sent back by student blaster, I was trusting my companion completely. For some reason though those two jobs causing the hamburger status were stopped at about 11%. Or at least PrimeNet claimed so. I have manually tracked them down and force fed them back to student blaster. He's chewing on these now before going back to his other work. I have no idea though how he lost his *.bu files for those two jobs either. I'm sure I didn't steal them - wouldn't be much use to me, now would they? So he had to start over again. I don't even know why he quit in the first place. Guess he just has a fetish for P-1. To each his own, I say.
Maybe I'll search his system for any old traces of the first time processing those two hamburgers.

Also the Tetris accelerators have done about 1'500 GHz-Days as of now. I just hope the home-calculators and student blaster won't get jealous. I still have to dive deeper into the Tetris world of things, but in any case their consolation prize is that they do calculations that my Tetris world won't as of now.

There are two other things I'm just going to abuse this cross conversation for. There is no way to claim an ANONYMOUS report, right? Even if the GUID of the device used to calculate it was the same as one placed in an account now?
And what is stopping me from signing up systems to other users and sending trash reports, or fiddle around with them in other ways? Not that I have such things in mind, my fellow companion would never allow me to do so anyway, but where is a leak there is an army of worms to dive into it.

[NBtarheel_33]

Originally Posted by Meral Harbes

There are two other things I'm just going to abuse this cross conversation for. There is no way to claim an ANONYMOUS report, right? Even if the GUID of the device used to calculate it was the same as one placed in an account now?

Try e-mailing George: woltman (AT) alum (DOT) mit (DOT) edu. Be sure to send along the results details.

And what is stopping me from signing up systems to other users and sending trash reports, or fiddle around with them in other ways? Not that I have such things in mind, my fellow companion would never allow me to do so anyway, but where is a leak there is an army of worms to dive into it.

I believe that the security hash (which makes use of the computer GUID and other pertinent data) thwarts this. And, of course, if you are bent enough to send garbage or violate the system, GIMPS is small enough of a community that someone will notice the erratic activity and take appropriate action. Please respect the integrity of the database, and refrain from "testing" PrimeNet or its mechanisms for the sake of frivolity. Legitimate questions or concerns should be brought to George's attention, rather than used as agents to prove a point.

Welcome to GIMPS, happy crunching, and good luck!

[Meral Harbes]

George had helped me with my request beautifully back then. Thanks George.

One other thing: Can I tell prime95 to keep updating the current results and send status information, but not automatically send new work requests? I'd like to finish my current set of jobs and slow things down after that.

[kladner]

Originally Posted by Meral Harbes

.......
One other thing: Can I tell prime95 to keep updating the current results and send status information, but not automatically send new work requests? I'd like to finish my current set of jobs and slow things down after that.

In prime.txt insert the line 'NoMoreWork=1'. Set it to zero if you want the program to get work.

Please Note! Stop Prime95 and shut it down before changing prime.txt.

[Meral Harbes]

Awesome, thanks!

Uh, nice. Is this an undocumented feature? =)

[kladner]

Many extra settings can be found in 'undoc.txt'. It can be very informative to check out all of the text files that come in the P95 package. Just don't mess with worktodo.txt unless you know what you are doing.

EDIT: .....and always shut down the program before editing prime.txt and local.txt.