Advanced search

Message boards : Number crunching : Work queue with latest 7.4.22 BOINC client

Author Message
Robert Gammon
Send message
Joined: 28 May 12
Posts: 63
Credit: 714,535,121
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 38430 - Posted: 12 Oct 2014 | 15:40:01 UTC

I have a SHOCKINGLY large number of GPU tasks to work through, Albert and Seti are the two monsters. Way way more than 24 hours of work.

GPUGrid did not respond with lots of work, however, Albert went wild with work that is due Tuesday.

Network buffer and additional work buffer were NEVER larger than 0.25 and I moved it down to 0.1 as projects were initializing after the latest Ubuntu crash. Now down to ZERO on Network Buffer, and 0.1 on additional work buffer.

Has anyone else seen something similiar with the Locutus of Borg BOINC client???

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 38461 - Posted: 13 Oct 2014 | 14:58:13 UTC - in response to Message 38430.

I have that happen with Primaboinca CPU WUs. For instance one client has a setting of .07 day work buffer and .17 additional work buffer. Right now it has 47 Primaboinca WUs queued and 1 Numberfields WU for a total of 90 hours of work. Five CPU cores are active so 90 / 5 = 18 hours of work when it should only be queueing 10 hours. This is a constant feature of Primaboinca but I haven't had it happen with other projects yet (I don't run Seti or Albert). Also haven't had the problem with GPU projects. Running Windows 7-64 7.4.22 BOINC client but it was also the same with earlier versions.

Robert Gammon
Send message
Joined: 28 May 12
Posts: 63
Credit: 714,535,121
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 38485 - Posted: 14 Oct 2014 | 8:16:58 UTC - in response to Message 38461.

So to solve my problem, I told GPUGrid NO CPU TASKS.

When I came in tonight, a Testmd app was running on 10 cores, shutting out almost ALL other work, and 17 hours to finish.

I run Linux EXCLUSIVELY.

Post to thread

Message boards : Number crunching : Work queue with latest 7.4.22 BOINC client

//