Message boards :
Number crunching :
Getting a lot of "trash" tasks that seem not to even be assigned to a GPU
Message board moderation
Author | Message |
---|---|
![]() ![]() ![]() Send message Joined: 27 May 99 Posts: 309 Credit: 70,759,933 RAC: 3 ![]() |
There are 9 GPUs, units are minutes Dev# WU count Avg and Std of avg GPU0 WUs:411 -Stats- Avg:1.5(1.37) GPU1 WUs:209 -Stats- Avg:2.7(0.86) GPU2 WUs:214 -Stats- Avg:2.6(0.92) GPU3 WUs:206 -Stats- Avg:2.7(0.96) GPU4 WUs:202 -Stats- Avg:2.8(0.97) GPU5 WUs:198 -Stats- Avg:2.9(0.93) GPU6 WUs:200 -Stats- Avg:2.9(0.95) GPU7 WUs:149 -Stats- Avg:3.8(1.22) ---- There are 3 GPUs, units are minutes Dev# WU count Avg and Std of avg GPU0 WUs:400 -Stats- Avg:1.2(0.79) GPU1 WUs:252 -Stats- Avg:1.3(0.38) GPU2 WUs:192 -Stats- Avg:1.8(0.51) I think something like this was mentioned in the past but I was unable to find a discussion about it. Was looking at my BoincTasks history file and discovered that the first gpu on both of my mining systems had a lot more tasks assigned to it than any of the other gpus. Looking at the raw history file there was no gpu assigned so the default when displaying was "0". I am guessing this is a bug in BT as the task was assigned but the polling time jumped over time it took for task to complete and BT missed the opportunity to get the device number. None of these tasks errored out. It just seems to be a waste of resources to get these tiny tasks. I will be happy to continue to crunch these small tasks but was wondering if they could somehow be bundled up to as to upload and download a "zip" full of them at once |
Ian&Steve C. ![]() Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 ![]() ![]() |
we can't control what tasks we are given. and there's no way to tell if a task is a "shorty/noise-bomb/etc" task until you process it. Boinc assigns tasks to whatever GPU is free. so if you have 9 GPUs and 8 of them are working on normal tasks that maybe take a few minutes each, then you have a bunch lined up that are short (only a few seconds to reach 30 signals), then that GPU will process, and finish, and be the only one available again to take the next short task. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours ![]() ![]() |
©2025 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.