Posts by Carl Christensen

1) Message boards : BOINC client : Tasks don't switch as scheduled... (Message 12014)
Posted 9 Aug 2007 by Carl Christensen
Post:
review things, rather than trying to change the already complicated & working fairly well scheduling on BOINC, we'll just have to bump up future boinc workunits to be years (I have just been using a default

<delay_bound> 30000000 </delay_bound>

which I put in the template ages ago, 30000000 being the number of seconds, which is about a year (347 days) --- it's just a number I came up with once long ago and has never been changed as there was always been 100 things more important to work on etc. so I'll tell Tolu & Milo to bump this up by a factor of 5 (I don't have anything to do with "day to day" CPDN anymore).
2) Message boards : The Lounge : Really, who likes the new BOINC logo? (Message 11966)
Posted 6 Aug 2007 by Carl Christensen
Post:
>>Until that time: Brrrr, David really doesn't have artistic feelings.

OK really, this is an absurd thing to say. I can vouch that David & I & Tolu from CPDN went to the Salvador Dali Museum in St Petersburg Florida (when we were all at the SC06 conference); and David was appreciating the art long after I was just waiting in the gift shop! :-)
3) Message boards : The Lounge : Eternity2.net, What Next? (Message 11893)
Posted 30 Jul 2007 by Carl Christensen
Post:
I agree that longevity should not be a factor -- in fact I wish/think some projects would have a definite end-point and know when to stop other than going on and wasting people's time; or have bona-fide new workunits and apps to keep things going. United Devices springs to mind; I know they have run their biochem simulations about 40 times each; a total waste of energy and the public's altruism. I'm afraid this will backfire eventually and volunteer computing looks like we don't have an "exit strategy" (like Bush & Cheney ;-)
4) Message boards : The Lounge : Really, who likes the new BOINC logo? (Message 11868)
Posted 29 Jul 2007 by Carl Christensen
Post:
coincidentally it seems to be Oxford's colors too! http://www.ox.ac.uk/
as well as the U of Washington: http://gohuskies.cstv.com/trads/020398aab.html
so it represents some boinc projects at least! :-)
5) Message boards : The Lounge : Eternity2.net, What Next? (Message 11815)
Posted 27 Jul 2007 by Carl Christensen
Post:
from the article in the Financial Times awhile back, it looks like the Sony PS3 efforts on BOINC or BOINC_style work on the PS3 will be "for profit." As long as participants know what they're getting into, I don't really have a problem with it (it would be nice if Sony or other "for profit" places put something back into BOINC & the community as IBM WCG has done).

I get a little more worried with "false advertising" or "false hopes" using BOINC, i.e. I personally feel a little uncomfortable with all the hype that some projects are "curing cancer @ home" etc, as the drug discovery process is quite far off from any BOINC biochem project (from what I've heard from friends in the pharm biz for example). At least SETI & Einstein & CPDN you get what they say -- try and find aliens or quasars or run climate models. I also get worried that there seem to be so many "senior project" or "grad student" projects jumping on the boinc bandwagon -- and all it takes is for one programming & security screwup to make everyone else look bad. But maybe I'm just paranoid! :-)
6) Message boards : The Lounge : Really, who likes the new BOINC logo? (Message 11814)
Posted 27 Jul 2007 by Carl Christensen
Post:
I kind of like it, it seems more professional & streamlined to me. When I make presentations on CPDN & BOINC, it always seemed a little goofy to have the old "jagged cartoon" BOINC logo ( example ). I think this new logo will fit in better with university & corporate logos on slides, presentations, websites, etc.
7) Message boards : The Lounge : Machines with 6-8GB of RAM getting more common? (Message 10839)
Posted 10 Jun 2007 by Carl Christensen
Post:
sending out multicore-ready models (via MPI) is something I've been toying with, esp for the higher-res models. although I'm not sure how BOINC will react to that, I suppose it couldn't stop our monitor job from launching a 2 or 4-CPU-usage model, but it would probably schedule other work to be done on these CPUs so we'd all be "fighting" for the same cores etc.

as a recap of the high-memory/res model I'm talking about -- at the 192 longitude x 145 latitude x 38 vertical levels (N96) resolution it seems to be using 4.7GB of RAM peak, typically at 4.5GB. It does a timestep (30 minutes of simulated time) in about 2 minutes! This is on a box that would do a normal 32-bit CPDN model (i.e. hadcm3) at a timestep in 2 seconds. So technically, this is what you would call a "doozy!" ;-)

the "normal" res version of this model (hadgam1a at 96 long x 73 lat x 38 vert levels) is quite a bit friendly -- running at 1.2GB peak RAM usage, 1GB typical, and a timestep every 15 seconds. All of these models are running at 64-bit maths (-i8 -r8) by the way -- and these numbers are for 64-bit machines & OS's (you can run -i8 -r8 on a 32-bit box/OS of course, and we probably will make versions for that; but the timings will be longer). We won't do 40 or 80 or 160 year runs with these models of course! It will be more like the Seasonal Attribution Project -- i.e. 6 months or a year or maybe two years simulated.

we can do a lot of science with the N48 ("normal") res version; although our proposal was really for the N96, so hopefully we can get at least about 1000 with big memory upgrades that don't mind using almost 5GB for CPDN! I guess we'll offer "extra credit" for that! ;-)

8) Message boards : The Lounge : Why I may be leaving BOINC (Message 10782)
Posted 8 Jun 2007 by Carl Christensen
Post:
it seems a bit silly to leave a project (or all projects) because of one nutcase posting. it's not like every kook on a board represents the entire project, scientists, participants etc! it looks like this person posted to a Japanese board, which doesn't have any moderation of the English-speaking/writing board.
9) Message boards : The Lounge : Machines with 6-8GB of RAM getting more common? (Message 10681)
Posted 5 Jun 2007 by Carl Christensen
Post:
I'm trying to get a feel for how you guys think home PC's are going, do you think in a year it will be common to see "off the shelf" PCs with 6 or 8 GB of RAM, or will it still be a rarity. Right now, it seems that 4GB is the typical max available, i.e. on Dell.com until you go to the "Big Business Servers" or whatever it still seems like their configs top out at 4GB.

I'm just wondering because I'm working on the latest-generation MetOffice climate model for CPDN (hadgam/hadgem). At the higher resolution (still lower than the Seasonal Attribution Expt but that is an old model), it likes to use 4.7GB of RAM! If we drop it to "usual" CPDN resolution of 96x73 (although with twice (38 vertical levels), it runs about 1.5-2GB, which is more manageable for today's higher end machines.

Of course the scientists want to keep the higher (192x145x38) resolution, but I'm wondering if even in a year we'd have 10 participants with machines that don't mind doing a job that uses 5GB of RAM, 2GB of disk space, 200MB uploads, 100MB downloads, etc!

PS -- In looking through the CPDN BOINC database of machine specs, it looks like there's only been about 1000 machines with > 4GB; out of a global / "ever tried CPDN" pool of 200K. So right now it still seems to be very uncommon; I guess just for specialist geeks etc.





Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.