mersenneforum.org  

Go Back   mersenneforum.org > New To GIMPS? Start Here! > Information & Answers

Reply
 
Thread Tools
Old 2010-07-14, 16:00   #1
Unregistered
 

5×311 Posts
Default Still modern?

Does it still make sense to do work in/for prime95? Or is it already too old, so the prime numbers in that are given out are already too small for a World Record/the program is slow compared to other newer ones....?

Just wanting to know if it still makes sense... ( I of course went as assignment for a World Record one on my 2.4 GHZ Dual Core (but jsut use one worker so I can do other stuff while the computer runs)...if I calculated it right the 49 Million Iterations will take me about 2 month?!?!, is that correct,does it make sense that I do it or should I rather do smaller assignments or nothing at all?

Demondog11
  Reply With Quote
Old 2010-07-14, 17:58   #2
Unregistered
 

4,451 Posts
Default

Also I heard I could get both of my Cores working on the same problem (number), how exactly do I do this? (I don't have a lot of knowledge about computers/mathmatics)
  Reply With Quote
Old 2010-07-14, 18:07   #3
Unregistered
 

BB16 Posts
Default

Sry for all my added stuff ;):

I figured out how to run multiple cores on one numbernwo, the problem I have right now is:

It automatically selects the one I had on my first core before (Worker 1). Is there a way, I can have it to use both cores on my second worker?
  Reply With Quote
Old 2010-07-14, 18:59   #4
NBtarheel_33
 
NBtarheel_33's Avatar
 
"Nathan"
Jul 2008
Maryland, USA

100010110112 Posts
Smile Welcome to GIMPS!

Quote:
Originally Posted by Unregistered View Post
Does it still make sense to do work in/for prime95?
Just FYI, the project that you are working for is called GIMPS, which stands for "Great Internet Mersenne Prime Search". Prime95 is simply a program that allows you (your computer) to complete and contribute units of work (e.g. trial factoring, LL tests, etc.) to the GIMPS project. So it is more correct to say that you're working on GIMPS, rather than working on Prime95.

Quote:
Or is it already too old, so the prime numbers in that are given out are already too small for a World Record/the program is slow compared to other newer ones....?
The author of Prime95, and director of the GIMPS project, George Woltman is constantly refining and releasing new upgrades to Prime95. Since the project started in early 1996, there have been some 25 different major releases of Prime95, with a 26th on the way in the near future. Each release takes advantage of state-of-the-art CPU architectures, instruction sets, etc. Moreover, the underlying code for the multiplication of large integers is the fastest of its kind in the world. For the primality testing of Mersenne numbers (GIMPS' primary goal), you literally cannot come close to anything as fast as Prime95.

The current world record prime is 2^43,112,609-1, found by GIMPS in April 2009 (see mersenne.org for more information). There is no proof (yet) that infinitely many Mersenne primes should exist, but it "makes sense" that there could be. So if you're looking to set a world record, you would need to test numbers of the form 2^p-1 for p > 43,112,609. You can accomplish this by choosing to receive the worktype "World Record Tests" in your Prime95 client. In fact, even if you select "First Time LL", there is a good chance you will receive an exponent big enough for a world record, should it be returned prime.

In summary, Prime95 is as new and as state-of-the-art as you could ever hope to get, and there is no danger of running out of world record test candidates probably within the lifetime of anybody currently alive!

Quote:
Just wanting to know if it still makes sense...
Of course. Unless someone comes along with a proof that there are finitely many Mersenne primes, the sky is the limit for testing. In theory, you cannot currently test any exponent higher than 596 million on Prime95, but for most, that's not a big deal, as testing such an exponent would literally take decades.

Quote:
( I of course went as assignment for a World Record one on my 2.4 GHZ Dual Core (but jsut use one worker so I can do other stuff while the computer runs)...
Actually, you can use both cores without *any* loss of performance. Another great thing about Prime95 is that it is designed to use only your computer's *idle* CPU cycles - in between when you do useful work. Consider that 2.4 GHz means that your computer is running at 2.4 billion cycles *per second*! Out of this huge number of available cycles, your work likely takes only a small fraction of them - the rest of the time, your system is just sitting there, idling. But with Prime95 running in the background constantly, there is never an idle cycle wasted...your computer's processor is running at 100% all the time.

Quote:
if I calculated it right the 49 Million Iterations will take me about 2 month?!?!, is that correct,does it make sense that I do it or should I rather do smaller assignments or nothing at all?
This is absolutely normal. 49 million is a common assignment length right now, and takes about 1 1/2 - 2 months on most hardware in use today. Remember that if you hope to find a world record prime, as I mentioned above, you have to test exponents bigger than 43,112,609. You can use your second core either to run another test, or run two cores on your single 49 million test - this would allow you to finish in roughly 60% of the time it would take a single core (say, 30-40 days or so).

If you don't fancy the idea of waiting so long for the LL test to complete, there are many other shorter assignment types available (none of which, unfortunately, would enable you to possibly find a new Mersenne prime). Trial factoring assignments take from mere minutes to hours, and double-checking assignments (which are desperately needed right now - see some of the other threads) take only a couple of weeks or so. P-1 factoring, which seeks to eliminate candidate Mersenne numbers with relatively small factors, is another good choice, if you have a decent amount of available RAM. P-1 assignments take from 24-72 hours or so.

You have a nice machine and an interest in the project, which are the two most important things to get started. Try different worktypes out and see what interests you the most. Eventually you'll even get used to waiting for an LL test to finish - a couple of months goes by faster than you might expect - and once you finish your first test, it's easy to be hooked!

Welcome to GIMPS, and please don't hesitate to ask questions here on the forum, if we can be of any more help. Your contributions to the project are definitely appreciated!
NBtarheel_33 is offline   Reply With Quote
Old 2010-07-14, 19:22   #5
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

23·7·167 Posts
Default

Quote:
Originally Posted by NBtarheel_33 View Post
In summary, Prime95 is as new and as state-of-the-art as you could ever hope to get, and there is no danger of running out of world record test candidates probably within the lifetime of anybody currently alive!
Thanks for posting that. I didn't have the time to do the long explanation. Yours is great.
Uncwilly is offline   Reply With Quote
Old 2010-07-15, 09:32   #6
NBtarheel_33
 
NBtarheel_33's Avatar
 
"Nathan"
Jul 2008
Maryland, USA

21338 Posts
Default

Quote:
Originally Posted by Unregistered View Post
Sry for all my added stuff ;):

I figured out how to run multiple cores on one numbernwo, the problem I have right now is:

It automatically selects the one I had on my first core before (Worker 1). Is there a way, I can have it to use both cores on my second worker?
I'm not sure if you've sorted out this question yet or not, but I'm having a little difficulty understanding what exactly you're asking. When you first signed up for GIMPS with a dual-core system, the default would have been for GIMPS to assign you *two* exponents to test - one for each core. Now, given that you have two cores and two exponents to test, here would be your options:

(1) Run one exponent on each core (this is actually the most efficient way to do things)

(2) Pick a single exponent and let both cores work on it (this will get the test done faster, but it will not be exactly twice as fast, due to the fact that the computation of an LL test cannot be completely parallelized)

(3) If you choose option (2), you will need to decide whether or not you wish to keep the second exponent that you were assigned. If you don't want to keep it, you can unreserve the exponent from your GIMPS account assignments page. On the other hand, you could just keep the other assignment in your worktodo file to start on once your first assignment finishes.

There is no way, however, to use both cores to test your first exponent, while simultaneously using both cores to test your second exponent. You can't have your cake and eat it too...
NBtarheel_33 is offline   Reply With Quote
Old 2010-07-15, 18:22   #7
henryzz
Just call me Henry
 
henryzz's Avatar
 
"David"
Sep 2007
Cambridge (GMT/BST)

2×2,909 Posts
Default

Quote:
Originally Posted by NBtarheel_33 View Post
Of course. Unless someone comes along with a proof that there are finitely many Mersenne primes, the sky is the limit for testing. In theory, you cannot currently test any exponent higher than 596 million on Prime95, but for most, that's not a big deal, as testing such an exponent would literally take decades.
Which i pretty certain would be contradicting an existing proof.
henryzz is online now   Reply With Quote
Old 2010-07-15, 19:36   #8
10metreh
 
10metreh's Avatar
 
Nov 2008

44228 Posts
Default

Quote:
Originally Posted by henryzz View Post
Which i pretty certain would be contradicting an existing proof.
No, the infinitude of Mersenne primes has not been proven.
10metreh is offline   Reply With Quote
Old 2010-07-16, 03:50   #9
joblack
 
joblack's Avatar
 
Oct 2008
n00bville

52×29 Posts
Default

Quote:
Originally Posted by 10metreh View Post
No, the infinitude of Mersenne primes has not been proven.
That's right but it's most probable that they are.
joblack is offline   Reply With Quote
Old 2010-07-17, 08:51   #10
Unregistered
 

7·727 Posts
Default

thanks a lot for the replies it helped me a lot! :)

I now was running 2 Cores on one problem, without any problems (both running at 100% speed), but when I changed, the type of assignment they where supposed to do they both just run at 50% (as if there would be 1 core on 100%). I read the Thread about that in the FAQ but it didn't help me :S. It definately is running just on 50% and it doesn't just display it as such (since the Temp is far from maximum). Is there a way to fix this?
  Reply With Quote
Old 2010-07-17, 09:08   #11
Unregistered
 

2·5·761 Posts
Default

Quote:
Originally Posted by NBtarheel_33 View Post

There is no way, however, to use both cores to test your first exponent, while simultaneously using both cores to test your second exponent. You can't have your cake and eat it too...
yeah I also didn't mean it like that. What I meant is that when I set the program so it would use both cores on one exponent. It automatically chose to use both cores on the first of the 2 exponents that were given to me, but I would have liked it better if it'd be working on the second one (since it already came further).

How would I unreserve exponents?

Thanks for all your answers, as you can see I'm not a computer (or math) guy...

Demondog11

P.s.: English also isn't my native tongue, therefore I want to apology for all the misunderstandings caused by my writing ;)
  Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Modern parameter choice for large 14e/small 15e projects VBCurtis Factoring 29 2016-02-12 20:45
Modern C Dubslow Programming 15 2016-01-12 09:13
Modern Foreign Languages in schools f1pokerspeed Soap Box 10 2013-12-28 16:27
Chance to use modern Graphics Cards as.. Marco Hardware 28 2003-11-02 23:21

All times are UTC. The time now is 23:04.

Fri Feb 26 23:04:36 UTC 2021 up 85 days, 19:15, 0 users, load averages: 2.96, 2.72, 2.63

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

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.