Participating in Distributed Computing projects that benefit humanity

Discussion in 'Politics' started by fspeculator, Jan 8, 2006.

  1. (I've searched around ET for posts on this subject, but couldn't find any. This is suprising, considering that us traders have modern, fast PCs which keep running 24/7, but only use them a few hours per day.)

    We all have probably heard about the legendary SETI@Home by Berkeley Univ (the mother of all distributed computing projects, searching for extra-terrestrial intelligence) which has over 5 MILLION users.

    The hottest project right now is Folding@Home, by Stanford Univ

    The DMOZ page http://dmoz.org/Computers/Computer_Science/Distributed_Computing/Projects/ lists over a 100 such projects, which request idle CPU time, for reasons ranging from finding a cure for cancer or AIDS, to building the latest CERN particle accelerator.

    Some are run by companies although it's not always clear at first, e.g. in the End User License Agreement you sign (online) to join Wold Community Grid, you'll see that you're actually signing an agreement with IBM.

    I've spent several hours trying to clear things up a bit for myself (as I wanted the best bang for the buck in performance, security and configurability) and I've summarized everything at:

    http://www.hyper.net/dc-howto.html

    If you are, or think about, donating your idle CPU time to projects that benefit humanity, you may want to read it.

    Happy New Year 2006.
     
  2. Actually, I've been running Folding@Homa via Google toolbar since 2004.

    But

    1/ G has stopped supporting F@H via its toolbar, new versions don't support it at all (says so in the URL you posted)

    2/ the code version supported via toolbar is the old TINKER code (vs GROMACS and even newer ones etc) which runs at snail pace

    and 3/ and perhaps most important, I'm not quite sure whether F@H is a project I should still be contributing to. e.g. Wikipedia's F@H page suggests that "recent developments have rendered it somewhat obsolete".

    I mention all this in my howto page. It took me quite a while to sort out all this stuff, which is why I went through the trouble of writing it down for others to benefit.
     
  3. If anyone else is actively contributing and have taken the time to look into them, I'd appreciate their feedback about which projects to contribute to.

    Some are more time-sensitive or prioritized then others, looking for computing power. I've joined most, but give out most to less widely followed ones (R@H, P@H etc) as everyone and his mother are already in F@H and S@H

    Btw I don't get all those people building "farms" (dozens of PCs at home) to work for SETI. As someone put it: "it (SETI) is a lot like the lottery. If you buy one ticket a week, you're a dreamer, if you buy 100 tickets a week, you've got a problem"
     
  4. Did you put that primer together (at hyper.net)? If so, nice job.

    I took a brief look at it, but I didn't see any explanation of the 'credits' system. At the BOINC site and others you can see running totals of how many credits different contributors have. Do you have a link to an explanation of this credit system? It's not like I would get involved in order to win a prize; I think it would be kind of cool to have my idle time working for a project I consider worthwhile; I'm just wondering. You also mentioned the idea of running 1 computer being like buying a lottery ticket...

    Here is the text of a PM I got from a friend - if you can't get the information you need, you may be able to contact this person at Anandtech and get something. You'd have to contact him (Rattledagger) by PM, I guess. This was in response to a question I asked about whether DC projects were secure. I described a situation in which the system might be taken advantage of

    "Yes, there have been people that "cheated" on some of the DC projects, most notably the Seti@Home project, both by returning hacked work units or duplicate caches of units, among others. The BOINC architecture is quite a bit more secure though, especially in regard to how credit is granted. If you search for posts by Rattledagger you'll find that he's more or less a BOINC "expert" - he could likely point you to the info you want on any aspect of BOINC. There are many different DC projects to choose from - a number of them with links can be found here..."

    I have no comment on the accuracy of these statements.

    http://forums.anandtech.com/messageview.aspx?catid=39&threadid=1770553&enterthread=y
     
  5. nitro

    nitro

    http://www.elitetrader.com/vb/showthread.php?s=&postid=783792&highlight=donate#post783792

    nitro
     
  6. TraderNik:

    Yes, I wrote it, thx.

    About the credit system, I don't care much about my "credits" (I never look at it, unless I want to make sure my PCs is actually processing WUs from project XYZ). I'd only care about whether the project is legit (my best measure would be citations by other researches).

    Credits are just a measure of PC speed, i.e. a benchmark (which is why these projects attract so many "overclockers")

    For me, credits are useful so I can tell that I don't need to run BOINC on that old Pentium-II/Win98 bec it's 20 times slower than a modern Pentium4/Athlon.

    BOINC has nice credit system:
    http://boinc.berkeley.edu/chart_list.php

    The non-BOINC projects are mostly run by commercial entities (IBM/WCG, grid.org/UD, etc) and their stats are too "grey" for my taste. One also gets no idea about the combined power donated to the project, unlike e.g. Stanford's F@H and UWashington's R@H which both report current TFlops estimate.
     
  7. What he describes is correct, but in the following sense:

    SETI (Search for ET intelligence) attracted HUGE number of following, about 5 MILLION. Incl. some ("wackos"?), which forgot that such a project is about the science

    So those wackos "cracked" the SETI client, to return bogus WUs, trying to run up their "credits" and get a better ranking. (bogus WUs could be due to faulty CPU, memory etc software bugs)

    This had an adverse effect, as Berkeley people started sending the same WUs to multiple people and only accepted the results if all received WUs matched.

    It was an issue, but only on the scientific part of the project. So some wacko is cheating some of their "credits". Like one could hack the arcade console and put a big high-score in PacMan. Big deal!

    In my opinion, as SETI was winding down (temp?) on the science front, Berkeley intentionally moved the platform to BOINC, so its HUGE userbase can now contribute to other programmes. They didn't say this, but I'm pretty sure that was why they went through this migration nighmare of moving SETI-Classic to BOINC.

    It was agreat move as far as I'm concerned.
     
  8. nitro, thx for feedback, I visited the links you posted.

    I think you'll find that my overview at

    http://www.hyper.net/dc-howto.html

    helps one understand how the various projects you cited, from climateprediction and SETI, to Folding and Rosetta and others fit together (from the perspective of the donor-user), so one can decide how to best join the worldwide DC grid to help his favorite "worthy cause".
     
  9. Actually it was someone else's comment:

    "it (SETI) is a lot like the lottery. If you buy one ticket a week, you're a dreamer, if you buy 100 tickets a week, you've got a problem"

    I think he was referring to those guys at the top-100 http://boinc.netsoft-online.com/rankings.php?list=r_expavg who run 400x P4 computers, but only contribute to JUST ONE project, whereas there are projects with more "tangible" results and/or are a pre-requisite for further research, like Human Proteome Folding.

    That's the only benefit I see with WCG vs BOINC, i.e. that WCG projects are decided by a board of advisors, as it seems many users will decide on factors like slick graphics...
     
    #10     Jan 11, 2006