| log in |
|
21)
Message boards :
Number crunching :
GPU Stock App Vs. SSE2
(Message 42540)
Posted 963 days ago by Astromancer. So I happened to wind up checking out the Op Apps on Arkayn's site and saw there was a SSE2 version of the GPU app. Will this give increased performance when compared to the stock app? And is it enough of a boost with a 3870 to bother doing it? Thanks guys! ::Edit:: I'd do the testing myself, but it's my parents computer that I slapped an old graphics card into. So not the easiest for testing! ;) |
|
22)
Message boards :
News :
windows intelx86 and x86_64 binaries added for the nbody simulations
(Message 41713)
Posted 1004 days ago by Astromancer. Arjan, thanks for that App_Info. It works great for running both at the same time. I've only wound up with one error so far and it seems that the WU is messed up (I'm the 2nd error). If anyone cares to look at the computer it's http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=201146. I have to admit I'm liking the short run times, but assume they'll have to unfortunately get longer to get any good data. :P My only real complaint is that it keeps on downloading 4 nbody WU's and only 2 ATI WU's. (0.02 day buffer just to keep from having to wait on downloads). But this seems to be more of a factor of for some reason BOINC thinks the ATI WU's will take like 4.5 hours. And it just so happens that it does an upload / download right before the task finishes so it works out perfectly. |
|
23)
Message boards :
Number crunching :
Amount of concurrent blocks
(Message 40543)
Posted 1069 days ago by Astromancer. The field concurrent blocks dictates how much of the GPU will be used by the application in a single iteration. The more concurrent blocks the faster the workunits are finished, but it will cause more user interface lag. Much like Sutaru I didn't see any difference in completion time or interface responsiveness between 128 and 50,000. And my screen is run through the video card. GTX260 55nm with 216 cores as well. Perhaps it would have more effect with single precision calculations where the gpu is more completely used for the computation? nmeofdst8, As for memory usage, GPU-Z reports the same ammt of memory used no matter what the setting is set to. |
|
24)
Message boards :
Number crunching :
Travis is camping with Milkyway
(Message 40191)
Posted 1084 days ago by Astromancer. He cursed the weekend when he asked us to not break things! hahaha :D |
|
25)
Message boards :
Number crunching :
Boinc 6.10.56 - backup project
(Message 40178)
Posted 1085 days ago by Astromancer. And on a completely unrelated note, I've gotten the nbody simulation code from another one of our astronomers here. And in a month or so we should have a single-precision GPU friendly application that will simulate the interactions between the milky way and other dwarf galaxies, which will be REALLY cool. Interesting... I had thought it was impossible to obtain the same calculation accuracy with single precision FP? I'm no programmer though. Is it some programming trickery or a complete redesign that doesn't need the same type of accuracy? Either way, very cool. Might be time to see if friends and family have any old graphics cards their willing to part with. :D I still have 2 free PCIe slots on my computers. |
|
26)
Message boards :
Number crunching :
GPU Requirements [OLD]
(Message 40147)
Posted 1086 days ago by Astromancer. CUDA might be more widely used, but the ATI cards are thrashing them in the WU completion category. True the ATI cards are faster in the projects they are used in, but this is the only project that I crunch that supports ATI cards currently. So I say NV for crunching (For myself at least) as it allows me to run more projects on my GPU as of right now. Hopefully OpenCL will change that at some point though! |
|
27)
Message boards :
Number crunching :
GPU Requirements [OLD]
(Message 40130)
Posted 1087 days ago by Astromancer.
Good to know for sure. Thinking of getting one though more for gaming than crunching. If I were buying for crunching I think I'd go NV again since CUDA is more widely used. But I've always liked the ATI drivers better :) |
|
28)
Message boards :
Number crunching :
Suspend work if CPU usage is above (Computing Preference Setting)
(Message 40128)
Posted 1087 days ago by Astromancer. Might be the wrong forum *shrug* But it seemed right. I just recently changed my settings on the MW website instead of the SETI one. And after having my crunching start randomly pausing I realized that by doing so it reset the "Suspend work if CPU usage is above" setting to 25% instead of the 0 (Ignore) that I had it set at in my SETI settings. So I figured I'd post this here figuring I'm not the only one who changes this setting. And I figure if we ask nicely that it'll get added when someone has some time. Thanks guys :) |
|
29)
Message boards :
Number crunching :
GPU Requirements [OLD]
(Message 40127)
Posted 1087 days ago by Astromancer. I am running an ATI HD Radeon 4870 X 2 Like TGG said, 6 WU's per CPU core. So if you want a bigger cache I guess you need an i7 for the HT. Think I need to change my cache settings so the SETI work cache is about the same size as the MW one... And to add to the GPU list, the 5830 is missing. Unless what I've read is wrong on it doesn't support double precision FP calculations? |
|
30)
Message boards :
Number crunching :
Boinc 6.10.56 - backup project
(Message 40110)
Posted 1087 days ago by Astromancer. Ah, thanks Paul. I may try going through BAM again to override the settings there, or turning the cache back on so I have some stored up WU's for those times where MW and Seti are having issues. Been getting a lot of project has no work when requesting GPU tasks from both of em in the past 5 days or so. |