BOINC Won't Take No for an Answer...
10 Jul 2015 19:52:11 UTC
Topic 198148
(moderation:
Greetings all -
I stopped Einstein a few months ago and decided to resume a few days ago. I usually minimize running in the summer due to the heat load in my office.
Upon installing the latest version of BOINC (7.4.42 - Win7 x64) and it downloading the latest apps, I immediately noticed all of the GW S6Bucket #2 tasks were aborting due to "Computational Error". I've been running this project since Feb of 2005 and never had a problem like this before.
Rather than hassle with trying to figure out what's wrong with GW S6 Part II, I decided to just not run that app. I selected on account preferences to not analyze these WUs, i.e., NO is selected for GW S6 #2.
The project is still downloading and running GW S6 WUs after updating and resetting the project.
How do you get BOINC to stop running GW S6 WUs ?
Many thanks...
BOINC Won't Take No for an Answer...
)
Regarding your large error rate, a prime suspect would be that you might find it to improve if reduced CPU clock rate, improved cooling, or both.
Regarding application restriction: did you check to see which location (aka venue) is assigned to the host in question, and that when you adjusted preferences you adjusted them specifically for that one of the four locations? (default, home, school, work). Alternately you can just adjust all four and not worry which is which.
I use the application restriction in Einstein preferences routinely, and it has consistently worked for me.
Thanks for the reply -
)
Thanks for the reply -
I've used the preferences as well on several occasions, but it seems to not be working. Maybe I'll try installing an older version...
As far as the errors are concerned, the processor is an AMD 9590 (navtive 4.72ghz) that's water-cooled. Working flat out on 8 WUs, well, before they error-out, the CPU never exceeds 50C. It ran 10 months 24/7 prior to this with no problems noted.
Again, thanks for the reply.
RE: Thanks for the reply -
)
Do you tried leaving a cpu core free just for your gpu to use? You said you run 8 cpu units and I noticed you also have an Nvidia 980 gpu in there too, leaving a cpu core free for your gpu could help with your errors.
As for how not to get certain units do you have both of these set to NO also?
Run beta/test application versions?
This helps us develop applications, but it may cause jobs to fail on your computer. no
Run CPU versions of applications for which GPU versions are available no
Hey Mikey, Yes, after
)
Hey Mikey,
Yes, after checking max temperatures using 8 cores, I changed preferences to 7 cores and got the same result. Also, I was able to isolate my system locking up at night to the S6 Bucket #2 WUs being processed.
So, here's what's happening: ALL S6 Bucket WUs are aborting due to computational errors after running for a variable time - which never happened before with the "normal" S6 WUs last fall, and the computer will eventually lock up when processing these as well. Last night while I was using my browser, the system locked up while I was moving the mouse. This doesn't happen with the S6 tasks suspended.
I'm wondering if some background task is interfering with the Einstein, maybe Norton Internet Security or something.
I'll try the Beta/Test and the selection of no CPU versions that have GPU.
Thanks -
RE: So, here's what's
)
I won´t claim to be an expert in the S6 and windows but it seems you are not able to complete S6 tasks at all.
I noticed this task
http://einsteinathome.org/task/507591968
which failed very quickly, with an error
[pre]
2015-07-08 18:15:37.2997 (4312) [normal]: FstatMethod used: 'DemodSSE'
2015-07-08 18:15:37.3007 (4312) [normal]: Reading input data ...
ERROR: illegal SFT-version (0 0 0 0 0 0 0 0) not within [1, 2]
ERROR: File-block 'h1_0430.20_S6GC1:0' is not a valid SFT!
XLAL Error - XLALSFTdataFind (/home/jenkins/workspace/workspace/EAH-GW-Release/SLAVE/MINGW32/TARGET/windows-x86/EinsteinAtHome/source/lalsuite/lalpulsar/src/SFTfileIO.c:292): Invalid data
XLALSFTdataFind() failed with xlalErrno = 137
[/pre]
and it reminded me of this thread http://einsteinathome.org/node/198138&nowrap=true#141721
I suspect something is (was) corrupting the downloaded files, and assuming a healthy disk, and file system, i would next look to Norton or other things which might tamper with downloads, and then resetting the project to clear any corrupted files.
On my linux system the files (like h1_0430.20_S6GC1) are all identical in size, 8445248 bytes long in the Einstein project directory.
Hope that helps.
RE: Hey Mikey, Yes, after
)
Have you set Norton to ignore the Boinc directories for it's scanning? If not do that as any REAL virus will try to infect other parts of your pc and get caught by Norton, but any 'false positive' in the Boinc directories will get ignored. If you use AVG you have to turn off the identity protection part or some projects will not work at all.