Message boards :
BOINC client :
GPU Resource Share Ignored
Message board moderation
Author | Message |
---|---|
Send message Joined: 30 Aug 05 Posts: 65 |
As I don't appear to be able to post to the alpha mailing list these days....not sure why, can't really be bothered working out why. I am running 6.10.17 (but I have seen this aver since GPUs have been detected by BOINC), 32bit XP SP3, Q9450, ATI4850 and ATI4870. AQUA(20%), MW (54%), Collatz (4%), SETI (12.5%, suspended), Spinhenge (9.5%, suspended). Connect every .1 days, cache 0.25 days. MW is set up to run 2 wu's per card at once while Collatz is at default of 1 per card. I understand that both MW and Collatz have been down and up a lot recently, but even so, BOINC is not respecting the resource share and is crunching too much Collatz and is more like 50/50. |
![]() Send message Joined: 29 Aug 05 Posts: 15483 ![]() |
Perhaps that everyone called Paul got banished from those lists. ;-) Ahem... can you please post a log that you ran with the correct flags? <cpu_sched_debug>: problems involving the choice of applications to run. <work_fetch_debug>: problems involving work fetch (which projects are asked for work, and how much). <rr_simulation>: problems involving jobs being run in high-priority mode. Other interesting debug flags: <debt_debug> <coproc_debug> <dcf_debug> There is no separate GPU resource share, you know that, I hope? There is only resource share, for the whole project, independent of what resources (CPU, GPU, magic, other) they use. |
Send message Joined: 30 Aug 05 Posts: 65 |
I'll set those tonight. Maybe I should have stated that BOINC is not respecting the resource share for the projects I am running on my GPUs. |
Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License,
Version 1.2 or any later version published by the Free Software Foundation.