1) Message boards : Application Code Discussion : CUDA for Milkyway@Home (Message 29502) Posted 19 Aug 2009 by Providence Christian School Post: For those of you worrying about the double precision requirement, there are other projects which do not require double precision accuracy and you can put your less well endowed cards on. (I don't have a CC 1.3 card.) Einstein - CUDA app in BETA, compute capability >= 1.1 Collatz - CUDA in beta, ATI in beta, the ATI app will even run on slow integrated graphics chipsets. Milkyway - CUDA in beta, compute capability >= 1.3; ATI app requires double precision capable card, (and some zippy SSEx optimized cpu apps) AQUA- CUDA app has been decommissioned, but they do have a multicore CPU app. GPUGRID - CUDA, compute capability >= 1.1 SETI - CUDA, compute capability >= 1.1 non BOINC: folding@home offers CUDA, ATI (needs a reasonable card to meet deadlines), multithreaded CPU and most of these also have optimized CPU apps which can use SSE(x) for significant throughput improvements. It seems like the projects which are working on GPU apps are also investing into SIMD based optimizations. So there are plenty of projects out there, one of which will certainly fit your GPU and CPU. For the geekiness of it, I run them all.

2) Message boards : Number crunching : MilkyWay_GPU (Message 20814) Posted 29 Apr 2009 by Providence Christian School Post: As MilkyWay_GPU is not listed as a project by BOINC, I can't sign up via my normal method "Signup for projects" from my BAM account page. I could attach to project directly instead of using the account manager, but will that cause problems later when I try to manage it with BAM? I could also create an account directly from the milkway_gpu page, but I still would have to attach directly to the project. So what is the best way to get signed up to get cuda gpu wus when they are ready?

3) Message boards : Number crunching : MilkyWay_GPU (Message 18858) Posted 15 Apr 2009 by Providence Christian School Post: Will the new GPU app Just Work without modfication of boinc or the milkway directory? Will it work like gpugrid and seti or do we need to treat it like an optimized app and go to zslip, dl new files and install them in the appropriate directories?

4) Message boards : Number crunching : Experiences with ATI app at Folding@home?? (Message 16826) Posted 25 Mar 2009 by Providence Christian School Post: I run F@H because it is the only distributed project which will use my low end 3450 graphics card. It does completely use a core, even after adjusting FLUSH_INTERVAL up, but the GPU produces a bit more than a core does and using the GPU is cool, so I left it. I hope they adjust the ATI app soon so that I doesn't waste a CPU core. They also need to adjust their deadlines because on this low end card, certain WU won't complete before the deadline.

5) Message boards : Number crunching : Lower spec ATI cards (Message 15019) Posted 12 Mar 2009 by Providence Christian School Post: That card should work on folding@home. They can use 2xxx, 3xxx and 4xxx ATI cards. I am in the same position. I have a 3450 and usually run BOINC, but to use that card I pull one core off of BOINC and run the folding@home gpu client. Click the checkbox to not run the larger units, but because you will have a hard time finishing a large WU within the deadline.

6) Message boards : Number crunching : New faster application? (Message 10423) Posted 12 Feb 2009 by Providence Christian School Post: The optimized apps rock. It thrills my geek heart to take advantage of more of the cpu capabilities. I have an AMD Phenom x3 with an ATI 3450 in it. Would the party who created the 64 bit win app for AMD/ATI GPU be interested in compiling a 32bit windows version? I assume that Brook libs will work on anything in the HD(2,3,4)xxx range (the same folding app does), so that it would also run on the 3450. At least I am willing to test it out.

7) Message boards : Number crunching : New faster application? (Message 9355) Posted 29 Jan 2009 by Providence Christian School Post: So have apps which take advantage of the SSE variants been recompiled with the updated code changes? URL? What would it take for the "stock" app to check for cpu capabilities and then download or use different code for the important parts? at least when it just comes to compiler switches. It seems silly to not use SSE(2,3,4) level instructions on machines which can take advantage of it and contribute a much greater benefit because of it. I enjoy running the various BOINC projects; I also enjoy squeezing every last bit out of my machines and taking advantage of all their capabilities. That is the premise of BOINC right? the primary reason my machines were purchased was not for BOINC, but I run BOINC to take advantage of all that unused capacity. The same optimization principles should apply to the BOINC apps, if the machine can do SSE3 then feed them an app which can use that unused capacity, rather than a lowest common denominator, at least for the major groups of machine classes. While discovering an optimized app for a project is fun, deploying it and maintaining it on a collection of machines is not efficient and if I can have a computer perform that repetitive task... I always find it puzzling that, cpu manufacturers tout cool new cpu features, which take months or years to make it into general compilers and then take even longer to make into applications... So your cool new whiz bang i7 is mostly running code limited to a 486 instruction set.

©2024 Astroinformatics Group