Welcome to MilkyWay@home

Posts by Martin Ryba

1) Message boards : MilkyWay@home Science : Billion-pixel Gaia camera to map (Milkyway) galaxy in 3D (Message 60762)
Posted 18 Jan 2014 by ProfileMartin Ryba
Post:
My quick take would be that it makes MW@H more relevant than ever. The Sloan Digital Sky Survey that they currently use is massive and nice but I imagine this will trump it by over an order of magnitude. The base data will be similar (star positions and magnitudes), just more. So, the algorithms and processing flow the fine folks at RPI have pioneered here would probably be the only tractable means for processing this new data set.
2) Message boards : News : Plot Progression! (Message 56419)
Posted 7 Dec 2012 by ProfileMartin Ryba
Post:
Do you have a link to the paper/preprint available? I'd love to read the details.

"Never ascribe to malice that which is adequately explained by incompetence." -Napoleon
3) Message boards : News : N-body updated to 0.2 (Message 43801)
Posted 13 Nov 2010 by ProfileMartin Ryba
Post:
Perusing the logs on one of my (old laptop) crunchers I encountered the message:
[quote]11/4/2010 12:47:22 PM Message from server: MilkyWay@Home N-Body Simulation needs 476.84 MB RAM but only 444.98 MB is available for use.
The N-body doesn't need that much memory. 15MB at most would be a high estimate for the 32k body workunits, which are the current largest.


It happened again, so I think there's an error in the build configuration for the N-body application that it tells BOINC client that it needs that much RAM.
Host Project Date Message
PamLaptop Milkyway@home 11/13/2010 9:12:31 AM Message from server: MilkyWay@Home N-Body Simulation needs 476.84 MB RAM but only 444.98 MB is available for use.

Marty
4) Message boards : News : N-body updated to 0.2 (Message 43653)
Posted 10 Nov 2010 by ProfileMartin Ryba
Post:
Perusing the logs on one of my (old laptop) crunchers I encountered the message:
11/4/2010 12:47:22 PM Message from server: MilkyWay@Home N-Body Simulation needs 476.84 MB RAM but only 444.98 MB is available for use.

So, it seems the N-body program is getting to be like Cosmology@Home in terms of being a memory hog. In some ways, I think the processing is similar. C@H maps how density fluctuations evolve in a potential field (that responds to the density fluctuations), M@H is mapping how stars evolve/move in a potential field that reacts to the star positions.

Comments?
5) Message boards : News : updated the nbody applications again (Message 42389)
Posted 27 Sep 2010 by ProfileMartin Ryba
Post:
Hi Travis,

For the (large) number of us running anonymous platform for the optimized apps for the original MW, can you make the application list link to the apps for download? Unless, can we go stock all the way now and get SSE2/SSE3 support on the main app? I had to hand-craft the URL to get to the download directory, unless I'm missing something due to the web front end redesign.

Thanks,
Marty
6) Message boards : Number crunching : wu 0.21 and Nbody simulations and file app_info (Message 42037)
Posted 11 Sep 2010 by ProfileMartin Ryba
Post:
Will just adding the above info to the app_info.xml cause it to automatically download the N-body executable from the MW server? I think it will, but I wanted to check before restarting the client.

Thanks,
Marty
7) Message boards : Number crunching : Increased RAM requirements? (Message 38772)
Posted 16 Apr 2010 by ProfileMartin Ryba
Post:

All the newly generated WUs should only ask for 50 MB of ram. Let me kick the work generator and see if that helps anything.

--Travis


That must have done the trick; at around 1AM my computer was able to get work; two hours earlier it had still gotten the error message. Thanks!

--Marty
8) Message boards : Number crunching : Increased RAM requirements? (Message 38699)
Posted 13 Apr 2010 by ProfileMartin Ryba
Post:
One of my machines is a somewhat older laptop that still does a reasonable job crunching, especially with the SSE2 optimized app. Lately though, its been reporting that it cannot run MW (even with the same app) due to a message:
Message from server: MilkyWay@Home needs 476.84 MB RAM but only 444.98 MB is available for use.


I had this problem with trying to run Cosmo, but given that's a real memory pig I'm OK with not running it. Was the RAM change on the server a conscious decision in preparation for the new app, or did something else happen?

TIA,
Marty
9) Message boards : MilkyWay@home Science : Congratulations to doctor Desell (Message 34832)
Posted 25 Dec 2009 by ProfileMartin Ryba
Post:

And there is one now at the NIF (took almost a year for all the security checks before he was allowed to ink his contract ;).


That's not surprising. Los Alamos does do an interesting mix of physics research with international collaborators (such as on NIF) along with super-secret nuclear weapons research (some NIF data feeds that as well, at least according to open sources). Does cause a bit of institutional schizophrenia. For a foreign national to get direct exposure to that environment must take a fair amount of background checking. The Department of Energy has a separate clearance process from the Department of Defense, but where we work people do have to wait many months to get some of the higher level accesses processed. I've had a clearance for nearly 20 years, so it's old hat to me by now. A bit of a culture shock to a recent academic though.
10) Message boards : MilkyWay@home Science : Congratulations to doctor Desell (Message 33424)
Posted 21 Nov 2009 by ProfileMartin Ryba
Post:
Second the congratulations! Welcome to the club! I know you have a postdoc through next summer...then the real challenge begins. I bailed and went corporate. Good luck Travis!
11) Message boards : Number crunching : GPU Projects (Message 31274)
Posted 21 Sep 2009 by ProfileMartin Ryba
Post:
FYI,

I don't have any GPU hardware so I'm fuzzy on the details, but Einstein@Home has a CUDA beta for at least the Arecibo binary pulsar survey. It doesn't need double precision. I think they're still working on a CUDA (and ATI) apps for the LIGO data.




©2021 Astroinformatics Group