Runtimes

Questions and Answers : GPU applications : Runtimes
Message board moderation

To post messages, you must log in.

AuthorMessage
Furlozza

Send message
Joined: 11 Sep 08
Posts: 14
Credit: 2,543,927
RAC: 0
Australia
Message 852855 - Posted: 13 Jan 2009, 2:08:30 UTC

Am running CUDA and it seems to be running okay, but have done three WUs and 1 has finished, one errored out and the other is growing in times, and sitting on 18 secs CPU time.

IS this normal or am I experiencing a build up to an error?
ID: 852855 · Report as offensive
BP1986

Send message
Joined: 24 Mar 06
Posts: 18
Credit: 88,495
RAC: 0
United States
Message 852976 - Posted: 13 Jan 2009, 8:16:09 UTC

What you are seeing is the CPU time used, not the GPU time used... CUDA uses a small amount of CPU to help it along, and BOINC records this time and GPU time is not recorded. Though in reality, you are doing MUCH more processing than 18 seconds.
ID: 852976 · Report as offensive
Profile Byron S Goodgame
Volunteer tester
Avatar

Send message
Joined: 16 Jan 06
Posts: 1145
Credit: 3,936,993
RAC: 0
United States
Message 852977 - Posted: 13 Jan 2009, 8:20:17 UTC - in response to Message 852855.  
Last modified: 13 Jan 2009, 8:23:23 UTC

Am running CUDA and it seems to be running okay, but have done three WUs and 1 has finished, one errored out and the other is growing in times, and sitting on 18 secs CPU time.

IS this normal or am I experiencing a build up to an error?


The task that is stuck at 18 seconds is probably a Very Low Angle Range task or VLAR (0.1 angle rangle and below) like the ones you have with a compute error, and you should abort any tasks like those. You should also reboot after having one.

It's also been suggested that you use the batch file lised in several of the threads so you can spot these tasks and abort them before they run. This thread has examples of those batch files.
ID: 852977 · Report as offensive
Furlozza

Send message
Joined: 11 Sep 08
Posts: 14
Credit: 2,543,927
RAC: 0
Australia
Message 853183 - Posted: 14 Jan 2009, 3:05:34 UTC - in response to Message 852977.  

Thanks Byron

Just ran it and "terminated" 10 WUs. Am using your amended version of the one that someone else quoted from someone else *grin*

SO, shall now sit back and see if this makes compute errors a partial or full thang of the past.

Thanks for the help.
ID: 853183 · Report as offensive
Maik

Send message
Joined: 15 May 99
Posts: 163
Credit: 9,208,555
RAC: 0
Germany
Message 853197 - Posted: 14 Jan 2009, 3:41:36 UTC
Last modified: 14 Jan 2009, 3:44:17 UTC

There is a update of a mod version from Raistmer
http://setiathome.berkeley.edu/forum_thread.php?id=51289&nowrap=true#853108
This will abort VLAR if it detects them and you dont have to do it manual ...

Before you install that make shure you have lastest gfx drivers ...
Set seti to NNT (no new tasks), wait until all tasks at your list are finished, uploaded and reported.
Then quit BM install the update, start BM and set seti to new tasks allowed ...
ID: 853197 · Report as offensive
Profile Byron S Goodgame
Volunteer tester
Avatar

Send message
Joined: 16 Jan 06
Posts: 1145
Credit: 3,936,993
RAC: 0
United States
Message 853198 - Posted: 14 Jan 2009, 3:42:33 UTC - in response to Message 853183.  

Thanks Byron

Just ran it and "terminated" 10 WUs. Am using your amended version of the one that someone else quoted from someone else *grin*

SO, shall now sit back and see if this makes compute errors a partial or full thang of the past.

Thanks for the help.


If you continue to have problems even with the script you can still try the lastest CUDA modified application V5a VLAR version

ID: 853198 · Report as offensive

Questions and Answers : GPU applications : Runtimes


 
©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.