Message boards :
Number crunching :
Take this CUDA or else!
Message board moderation
Author | Message |
---|---|
Mr. Hankey Send message Joined: 8 Jun 99 Posts: 30 Credit: 34,491,352 RAC: 0 |
So I finally decided to get CUDA going on my one computer that could do it. I got it all set up using the various optimized apps and boinc 6.6.18. So now I am crunching AP5.0,AP5.03, MB6.03, MB6.08. I then decided that I would try attaching my GPU to CPUGRID. I thought boinc would switch the CUDA processing between seti and cpugrid. The very first cpugrid unit errored out so I figured I would try not running both CUDA projects at the same time. Fine, I went to the seti prefs and disabled use GPU so that I wouldn't get anymore CUDA work and I could clear out my CUDA queue. Problem seems to be that I do continue to be sent CUDA WUs. How do I turn the spigot off? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Fine, I went to the seti prefs and disabled use GPU so that I wouldn't get anymore CUDA work and I could clear out my CUDA queue. Problem seems to be that I do continue to be sent CUDA WUs. How do I turn the spigot off? We rather think (but it's no more than a guess) that when you present yourself at the server with an app_info and a BOINC that between them say 'I have a CUDA app and I want some CUDA work', then the server takes you at you word and gives you what you ask for. (Whatever the web preferences say). I agree, that sounds different from the behaviour with respect to Astropulse/Enhanced tasks, where web preferences do seem to affect what work gets issued, but hey, I didn't write this stuff. So I think you're stuck with the usual ways of running down a cache: set "No New Tasks" or a reduced cache duration. If you want to selectively run down the CUDA cache, opt to crunch Astropulse only for the duration. And when you've run the CUDA cache dry, remove those entries from app_info.xml |
Mr. Hankey Send message Joined: 8 Jun 99 Posts: 30 Credit: 34,491,352 RAC: 0 |
I will give that route a try. Thanks. Am I correct though that two different projects cannot use the GPU and BOINC will manage the switch back and forth between the two? |
MarkJ Send message Joined: 17 Feb 08 Posts: 1139 Credit: 80,854,192 RAC: 5 |
So I finally decided to get CUDA going on my one computer that could do it. I got it all set up using the various optimized apps and boinc 6.6.18. So now I am crunching AP5.0,AP5.03, MB6.03, MB6.08. You should be able to do Seti (all varieties) and GPUgrid. I do on all my cuda-capable machines. With cuda it processes the tasks in deadline order. GPUgrid currently runs 4 day deadlines, so basically they get in before the Seti work does. Eventually the Seti work will get a look in. GPUgrid requires a fairly grunty card, and a fairly up to date one, the faster the better. Seti doesn't use as much of the card as GPUgrid so lesser cards can be used here. What model card do you have and how much memory has it got? BOINC blog |
Mr. Hankey Send message Joined: 8 Jun 99 Posts: 30 Credit: 34,491,352 RAC: 0 |
I am using an 8800 GT with 512 MB |
©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.