10/22   What are some applications or general set of problems that are most
        suited to be solved on server farms?  I know of stuff like SETI.
        But I was thinking along the lines of problems that are specially
        suited for large number of machines interconnected to massive
        storage in one centralized place (little network latency).  Thanks.
        \_ hosting large web sites.
        \_ lots of bioinformatics problems
        \_ http://www.fightaidsathome.org  BTW I vaguely remember that its number of
           participants is much lower than SETI@home, even though fighting AIDS
           seems more important.  Anyone knows why?
           \_ because if we find aliens it'll change the course of human
              history dramatically, but if we stop aids, it just means one
              less way out of thousands to die.  death is common, aliens are
              not.  others understand that fighting aids is not more important
              when most aids carriers got it through voluntary activities.  the
              only true aids victims are hemophiliacs and surgey patients who
              got tainted blood and similar situations.
              \_ bad troll--the true nutcase AIDS troll is Duesberg's
                 "AIDS doesn't exist" position.
                 \_ it's not a troll.  it's the truth.  i expressed my opinion.
                    if you disagree, say why.  if you take it up the ass a few
                    times a week from strangers you're going to get aids and
                    that's no one's fault but your own.  i've got no sympathy
                    for any new cases that have come along in the last 10 years
                    from people engaging in known risky activity.  the same
                    goes for iv drug users, or anyone else engaging in high
                    risk activity that has been *known* to be high risk and a
                    near guarantee of a death sentence since the early 90s. at
                    this point no adult has the excuse of not knowing what
                    causes aids or how to avoid it.
                    \_ seti@home was first, and had great publicity.
                       I'm guessing if the roles of the two projects
                       were reversed in terms of publicity and timing,
                       it would be the other way round.
                    \_ What about the third world?
              \_ I agree on the true victim part.
        \_ I guess the class of problems are those which the ratio of CPU time
           to data size is high, and the data can be divided into independent
           sets.
        \_ molecular biology, specifically for studying protein folding.
        \_ Oceanography, meteorology, and lots of other Earth sciences. We
           use supercomputers for gravity modeling. The best problems are
           those where the next operation does not involve the result from the
           previous operation (e.g. Monte Carlo sims). --dim
           \_ Yes. a variety of statistical analyses and modeling is most
              common.  MCMC, especially, since you can run 100 independent
              threads, and then merge them all together for analysis. -nivra
        \_ Neuroimaging.  Structural MRI comparisons across subjects and
           through populations using volumetric tensor fields.  Functional
           studies making comparisons across groups for 4D scans.  Check out
           http://www.loni.ucla.edu  -drex