Seti@Home optimized science apps and information
Optimized Seti@Home apps => Windows => Topic started by: sunu on 15 Aug 2007, 12:38:36 pm
-
With the new 2.4 Win32 Chicken soup I'm seeing very high duration workunits that end up with result overflows.
Usually workunits with result overflows are discarded after a few seconds/minutes of computation. Now I am seeing many workunits with 15000+ sec computation times ending with result overflows giving me only 0.x credits.
Check this workunit http://setiathome.berkeley.edu/workunit.php?wuid=147539615
If you see both results, we are both using Chicken optimized apps. Mine is the P4 2.8 Ghz. The second result is from a C2D E6600 and it took him almost 11000 sec for this workunit and we ended up getting 0.43 credits!!!
Is there something wrong with the new apps?
I have many others such as this one that are waiting validation. Check this http://setiathome.berkeley.edu/results.php?hostid=3690316&offset=20
-
Joe Segur has been looking into something similar for us on the main SETI board:
http://setiathome.berkeley.edu/forum_thread.php?id=41541&nowrap=true#619629
If your WU data has the same characteristic - negative <triplet_thresh> - then it's a splitter problem, not an application problem.
Joe's extremely good advice is to suspend any such workunits, until the project can cancel them: if you abort them, they'll just get re-issued to some other poor unsuspecting sucker!
-
I also just saw that thread. This isn't a chicken thing.
Well, we just have to wait for the little quirks of multibeam workunits to get ironed out. :)
-
Nothing but lots of frustration with these workunits :)