Computer Specs

i have access to a computer with 1gb of ram and runs at 3.8ghz,would this be suited to finding new mersenne primes,what are the specs of some of your computers?sorry but i dont have any other specs at the minute,if there are any others you would like to know to help me im sure i could get them for you.

Thanks and happy xmas

[petrw1]

Originally Posted by Unregistered

i have access to a computer with 1gb of ram and runs at 3.8ghz,would this be suited to finding new mersenne primes,what are the specs of some of your computers?sorry but i dont have any other specs at the minute,if there are any others you would like to know to help me im sure i could get them for you.

Thanks and happy xmas

Those are pretty decent specs. You'd be in the top half I'm sure.
If you check the benchmark page on the www.mersenne.org site (once it is back up) you will see hundreds of submitted benchmarks that include the corresponding computer specs.

I have a few computers ranging from a PIII at 0.866 Ghz to a Quad core each rated at 2.87 Ghz.

[henryzz]

your pc seems to have a fast single core cpu(single core is a presumtion but if that presumtion was false then you have one of the best pcs) which means it will only do one test at once but will do it faster than a lot of multicore cpus that almost always run at a slower speed per core
have you already downloaded Prime95 already
if so could you run a benchmark and post it here i would be interested to know how fast your pc runs

all help is welcome to GIMPS
i think computers a tenth of the power of your pc could still help in small ways(although the processing/power consumtion ratio would be a lot worse)

one of the most common pcs on the forum currently is a Q6600 which is a quad core pc running at 2.4 Ghz although it seems people are starting to buy slightly better things more

edit: a 3.8 Ghz cpu might have hyperthreading which would mean you appear to have 2 cores

I looked at the system info when you go into my comp,thats where i got my info from,where can you get other specs?it says 4cpu in the system info aswell,i presume that means its a quad core.i could also possibly get access to two even faster comps from work,(all 3 are used for engineering purposes).havent got prime95 on them yet but i might set it up in the new year,the 1 i talked about earlier doesnt have internet access so im sure it'l take a bit more time to setup.you can test the same number on several different comps,right?do they all have to be different tests or could i run different stages of the ll test on different comps at the same time?

Thanks

[cheesehead]

Answers to your last two questions:

Originally Posted by Unregistered

you can test the same number on several different comps,right?

Not on different CPUs or computers at the same time (or -- if you did, it would be useless duplication). GIMPS's assignments aren't like most other distributed computing projects -- each GIMPS L-L assignment (work unit) consists of long calculations done sequentially, which may take weeks or even months to complete.

(It's possible to stop work on a particular number in the middle of an assignment, transfer that number's save file (status file) to another computer, then resume work on that number on the second computer from the point where the first one stopped -- but I don't think that's what you're asking about.)

However, what you _can_ do is get more than one assignment at a time, for running on separate CPUs of your computer. You _can_ run up to four different assignments at the same time on your four-CPU computer. OTOH, I recommend starting with just one or two, to see what's it's like, then get two more assignments later once you have the feel of it and the first one or two are running smoothly.

could i run different stages of the ll test on different comps at the same time?

No. And it's not because GIMPS made some decision not to allow that -- it's because of the inherent mathematical nature of the L-L test. The L-L test has to be done sequentially and completely, start to finish, to get the result which you'll then report.

(BTW, no credit is given for partially-completed assignments. Credit is only awarded after you make your assignment completion report to GIMPS.)

[S485122]

Originally Posted by Unregistered

It says 4cpu in the system info as well, I presume that means its a quad core. I could also possibly get access to two even faster comps from work (all 3 are used for engineering purposes).

I am almost sure that computer is a server or a workstation with 4 Xeon CPU's, not a desktop with a QuadCore. If that is the case forget about overclocking.

I do not want to put a damper on your enthusiasm but you must ensure you get permission to install Prime95 on those computers, otherwise you could get in serious trouble. From the readme.txt file

If you are running this program at your place of employment, you must
first GET PERMISSION from your network administrator, boss, or both.
This is especially true if you are installing the software on several machines.
Many companies have policies that prohibit running unauthorized software.
Violating that policy could result in termination and/or prosecution.

Jacob

[Freightyard]

If it's an Intel CPU, this utility will tell you EXACTLY which one it is:

Intel Processor Identification Utility
http://support.intel.com/support/processors/tools/piu/


This program claims to identify AMD processors:
http://www.softpedia.com/get/System/...entifier.shtml

how can i overclock my comp?il put up the figures from the benchmarking sometime tommorrow hopefully.

[cheesehead]

Originally Posted by Unregistered

how can i overclock my comp?

Please don't. Overclocking greatly increases the chance of execution errors that could render your whole L-L test useless. The increased heat it generates may reduce your CPU's lifespan. (As Jacob reminds us -- is it your computer that you might damage, or does it belong to your employer or someone else? If you're not the owner, do you have written permission from the owner to (a) install prime95, or (b) to overclock it?)

Balance those factors against the time savings that overclocking will bring. 10%? 15%?

These numbers are BIG, and it's understandable that one can get impatient when the computation progress c_r_a_w_l_s along. But accuracy of the result is the most important consideration. There were years of effort put into making prime95/mprime software as fast as possible without sacrificing accuracy. One circuit glitch in the middle of your computations can ruin the eventual result, and it may be a few years until a doublecheck discovers that your result was erroneous and has to be done all over again by someone else.

[joblack]

Well, overclocking isn´t bad per se.

After you overclock try that the torture test from Prime95 for a decent period of time (at least several hours) - just checking if windows works isn´t enough.

My quad core works fine - but you have to finetune it.

Originally Posted by cheesehead

Please don't. Overclocking greatly increases the chance of execution errors that could render your whole L-L test useless. The increased heat it generates may reduce your CPU's lifespan. (As Jacob reminds us -- is it your computer that you might damage, or does it belong to your employer or someone else? If you're not the owner, do you have written permission from the owner to (a) install prime95, or (b) to overclock it?)

Balance those factors against the time savings that overclocking will bring. 10%? 15%?

Overclock can bring more than that ... my quad jumped from 2,4 ghz to 3,0 ghz

[Brian-E]

It depends how you look at it, doesn't it?

A certain amount of overclocking probably increases your own individual chance of finding a new Mersenne Prime because you will be able to test more Mersenne numbers. But the resulting decreased reliability of your tests is negative for the GIMPS project as a whole. One scenario, to serve only as an illustration, is that a mersenne prime is missed on its first LL-test because an overclocked machine gives a false-negative result. More typically, false results turned in by overclocked machines lead to mersenne numbers needing to be tested a third time when the double check (carried out a few years after the first LL-test) shows the discrepancy. That is a waste of everyone's resources.