Message boards :
Number crunching :
Type of work units v credit
Message board moderation
Author | Message |
---|---|
David Steel Send message Joined: 12 Jul 01 Posts: 52 Credit: 17,713,386 RAC: 0 |
On my ATI PC I'm getting long and short WU's and also AstroPulse, the run times and credit for each type of WU is similar, EG: longs take 1000 to 1300 seconds for 100 - 140 credit. But what's with these .vlar work units? They take twice as long for the same or less credit? Can I change a setting so I don't get these? Thanks |
Area 51 Send message Joined: 31 Jan 04 Posts: 965 Credit: 42,193,520 RAC: 0 |
On my ATI PC I'm getting long and short WU's and also AstroPulse, the run times and credit for each type of WU is similar, EG: longs take 1000 to 1300 seconds for 100 - 140 credit. .err no, part of the rich tapestry I'm afraid. |
.clair. Send message Joined: 4 Nov 04 Posts: 1300 Credit: 55,390,408 RAC: 69 |
vlar on a GPU is a feature of ATI yumyness :¬) |
David Steel Send message Joined: 12 Jul 01 Posts: 52 Credit: 17,713,386 RAC: 0 |
Soooo, ATI users get shafted on credit but NVidia users don't? (I use both by the way) Does anyone know why .vlar's give crap credit? |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
The kitties crunch 'em all....... Damn the credits......full speed ahead. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
Nvidia took the initiative, and helped work out the first GPU apps for Seti. ATI, whatever, did not. After that, the Lunatics developers took over and did even better. NV has the upper hand........and might well always be the preferred card on Seti. Get over it. The 680 and the 690 are doing very well, and Jason shall see to it that they prevail. They have more to offer than the ATI camp does. Simply put. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
Soooo, Because the telescope was pointing at one spot for the entire duration of the WU, all the data is from one remote target. There are consequently fewer analyses to do but each has maximum data, just the kind of thing which is more difficult to process in parallel fashion on GPU. For nVidia GPUs and the application port to CUDA which nVidia provided it proved so difficult that eventually they were excluded from being allowed to participate in that kind of work. Even for CPU crunching some systems struggle with those, see The Attack of the Killer 58.7s for instance. (That predates the shift to finer chirp resolution, if credits were still being calculated the same as then VLARs would now be getting 117.4 credits.) BOINC's basic measure of how much work is in a WU comes from the splitters as rsc_fpops_est which was roughly calibrated for typical CPUs in early 2008 based on the Estimates and Deadlines revisited thread. Perhaps that will be adjusted to give more credit for VLAR work before SETI@home v7 is released here, if the OpenCL S@H v7 ATi applications become stock releases. Joe |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.