e-mail to supercomputer owners
 

Well, in the light on the recent SC 2004 conference, I am planning to send this e-mail to the groups that own a powerful supercomputer. It's far from perfect, so do you guys have any suggestions?

subject:
[quote]An invitation to help find large Mersenne primes[/quote]
message:
[quote]To whom it may concern:

My name is Danny Chia, and I am a freshman at UC Berkeley majoring in Engineering Mathematics and Statistics. I would like to congratulate you on the success of your <name> supercomputer during the SC 2004 conference and being the <rank> most powerful supercomputer in the world, according to top500.org.

Such a computer could have a large positive impact on distributed computing projects. In my spare time, my personal computer runs software provided by GIMPS (Great Internet Mersenne Prime Search) [mersenne.org] that tests numbers for primality or trial factors numbers for compositeness.Although our project has a fairly large throughput (over 16 t-flops), a contribution of resources by a supercomputer would help us even more. So I am asking you to consider using your supercomputer to help find such prime numbers.

Such engineering wonders can greatly help us expand our knowledge of number theory.

Thank you for taking the time to read this.

- Danny[/quote]

dont forget to add there is a prize possibily and
something else idk like test your supercomputers power and see how much it adds to gimps network.

It's good except you need a space after "compositeness" and tell them it only uses spare cycles.

Here is a slightly revised version. Any other suggestions? Thanks for the input, guys. :smile:

[quote]To whom it may concern:

My name is Danny Chia, and I am a freshman at UC Berkeley majoring in Engineering Mathematics and Statistics. I am writing to congratulate you on the success of your <name> supercomputer during the SC 2004 conference and being the <rank> most powerful supercomputer in the world, according to top500.org.

Such a computer could have a large positive impact on distributed computing projects. In my spare time, my personal computer runs software provided by GIMPS (Great Internet Mersenne Prime Search) [mersenne.org] that tests Mersenne numbers (numbers of the form 2^p-1, with p being prime) for primality or trial factors them for compositeness. The program uses spare CPU cycles to make calculations. Using this concept, we have discovered several record-size prime numbers, such as 2^24036583-1, which is over 7.3 million digits long. Although our project has a fairly large throughput (over 16 t-flops), a contribution of resources by a supercomputer would help us even more. So I am asking you to consider using your supercomputer to help find such prime numbers.

Aside from finding prime numbers, various people have been using distributed computing projects to take part in the RSA number factoring contest, or to try to find a factor of the double-Mersenne number MM61 (2^(2^61-1)-1).

Such engineering wonders can greatly help us expand our knowledge of number theory.

Thank you for taking the time to read this.

- Danny[/quote]

hopefully they will get them most of the time there deleted by the people that read the other peoples email they do that to my teacher at school read tehre emails delete the ones asking then to conseder running prime 95 on school comps

This is bordering on spam. I would be inclined not to send any such email. If you do decide to go ahead and send an email like this, make clear that you do not have an endorsement from the founders of the GIMPS project to send such an email and that this is your own idea.

We do not want George to be flooded with angry emails from people upset over this email.
Thanks

It's a bad idea, ixfd64. No supercomputer owner or administrator would welcome this begging for free time by a distributed computing project that's doing just fine on its own.

At the very minimum, as Garo wrote, you have an obligation to add a prominent disclaimer that you are not authorized by GIMPS to send this message.

But please don't send it at all. [b]It can only cause trouble. It cannot help, but may very well [u]hurt[/u], GIMPS.[/b] Don't do it.

So is there no way to convince them to find large prime numbers? Or is there a better method than sending such an e-mail?

[QUOTE=ixfd64]So is there no way to convince them to find large prime numbers? Or is there a better method than sending such an e-mail?[/QUOTE]

Think about this from their point of view. What would they get out of it? A press release that says their multi-million dollar computer can do what a network of unpaid volunteers is doing for free? "As good as free" doesn't keep the funders happy. Suppose they find a prime - then it's worse because the news reports compare them to the single computer that found the last prime. Suppose they find a 10 million digit prime - now the news compares the prize fund to their electricity bill.

There is a long tradition of using idle time on large computers for number theory. For example, read the introductions to the Cunningham book. GIMPS, NFSNET, and ECMNET are all reasonable projects as idle time fillers that could run on individual processors. But modern processors often go into low energy states when idle, so there might not be much use for that.

William

[QUOTE=ixfd64]So is there no way to convince them to find large prime numbers? Or is there a better method than sending such an e-mail?[/QUOTE]
I've kept out of this thread up to now as I don't like to discourage enthusiasm. Call me a coward if you wish, but I'm in agreement with the nay-sayers. I believe that, at best, such an email will be useless and could cause damage.

Most owners of supercomputers, as far as I am aware, are oversubscribed when it comes to computational projects. Many are well aware of the existence of such as GIMPS, zeta grid and SETI@home. They also are well aware that although supercomputers are able to participate in such projects, rather more ordinary computers [i]en masse[/i] are also extremely capable. Quite rightly, they give much higher priority to projects that can not run cost-effectively on more common machines and which actually require a supercomputer to run at all.

As I said, I don't wish to discourage enthusiasm. If you could persuade a large institution with thousands of perfectly ordinary machines to contribute you would advance the GIMPS cause much more and, IMO, be more likely to succeed in your attempt. There are many such organizations in both public life and in corporations.


Paul

[QUOTE=xilman]As I said, I don't wish to discourage enthusiasm. If you could persuade a large institution with thousands of perfectly ordinary machines to contribute you would advance the GIMPS cause much more and, IMO, be more likely to succeed in your attempt. There are many such organizations in both public life and in corporations.[/QUOTE]

If I could get my employer to run Prime95 on all the PC's that it owns, we would likely be in the top 10 producers (top 100 within 1 month likely). There are many similar cases that could provide a huge number of machines. I suspect that many would prefer to run folding@home or the like, rather than Prime95.