Martin wrote: > Thanks for some initial answers. > > And yes, the question was deliberately vague to see what type of > significance was argued. > > > Ian Halliday wrote: > >> Martin has probably not come to the most unbiased place for an answer >> to his question, as I would expect most people here to believe that >> GIMPS is indeed significant. I don't know of any other mathematical >> search doing as much reliable work. >> > > That in itself is significant, if only for the world of distributed > computing. > > > >> [EMAIL PROTECTED] used to be popular, but I haven't heard anything about it >> for a while. It had many more participants, probably because looking >> for extra-terrestrial life is sexier than looking for massive prime >> numbers. However, I understand that their search lost popularity when >> the server ran out of source data so kept on recycling already checked >> data. >> > > That was a long time ago for a brief period during their change-over to > the Boinc system. A big advantage of Boinc is that if there ain't work, > then none gets sent out. However, [EMAIL PROTECTED] has a wealth of data that > so far > will continue to outstrip processing capability for a good while yet. > (Provided Arecibo is kept funded somehow and that [EMAIL PROTECTED] somehow > keeps > itself nefariously funded.) > I split my spare cpu time between here (GIMPS) and [EMAIL PROTECTED] on BOINC. Both of them keep my cpu's quite busy. Some of the other BOINC projects available include protein folding, drug research, and a couple of number theory experiments vaguely similar to GIMPS.
> > >> Google apparently offers something similar, related to protein folding >> as a possible way to assist cancer research. I don't know how much CPU >> power they have at their disposal. I believe Amazon are also either >> looking at or have already implemented a similar project for their >> users. >> > > Interesting... Why that bias? More 'sexy' subjects or corporate sponsored? > A little bit of both, I'd say. However, I have philosophical problems with donating my spare cpu time for free to a for-profit company, even if the end result is better medical care. If they were to promise to cut their price, or forego patent protection, or some other kind of repayment, of any resulting products, that might sway me back the other way, though. > > >> GIMPS currently harnesses power equivalent to over 1000 CRAY >> > [...] > >> 1996. Also, millions of candidates have been eliminated by being shown >> to be composite, 118 of them by me. >> >> Certainly GIMPS is significant, but I cannot comment on how it >> compares with the others. It would be churlish to say that we have had >> a result ten times but SETI haven't had any... >> In that kind of research, negative results are still useful results, in that they can help rule out certain ETI theories, based on the known receiver sensitivities and search distributions that have been done. >> Ian >> > > Thanks, good summary. > > > My main thought is whether we have reached the point of diminishing > returns as the primes become more scarce more quickly than compute power > increases to find them? > > Are there other scientifically interesting searches that should be > promoted until we get refined algorithms, new strategies, or x1000 > faster hardware to push GIMPS further anew? > > > So, my main question is: Have we reached the point of diminishing > returns for GIMPS? > Probably, but I don't know if the reduction in frequency of primes is counter-acting the rise in cpu power or not. It might be an interesting study, though... D _______________________________________________ Prime mailing list [email protected] http://hogranch.com/mailman/listinfo/prime
