Invalid results on N-Body Simulation process
log in

Advanced search

Message boards : Application Code Discussion : Invalid results on N-Body Simulation process

Author Message
Sonoraguy
Send message
Joined: 18 Feb 13
Posts: 1
Credit: 261,599
RAC: 0

Message 58165 - Posted: 7 May 2013, 22:59:31 UTC

Hello,

For the second time in a week, I've had a process that ran for 50,000+ seconds that came up "Invalid" when none of the other results completed and - according to the site - expended 0 or close to 0 seconds which seems like they hardly got started.

So.... if the result were compared to something else that completed, I could see that but it feels like my compute cycles are being wasted here.

On World Community Grid, they compare completed results ONLY until they have a quorum. In 7,000 processes run, I haven't had one Invalid BUT I really like the idea of contributing to mapping our galaxy.

So.... can we do a better job of utilizing the contributions being made?
____________

mikey
Avatar
Send message
Joined: 8 May 09
Posts: 2033
Credit: 180,983,475
RAC: 310,209

Message 58170 - Posted: 8 May 2013, 10:46:23 UTC - in response to Message 58165.

Hello,

For the second time in a week, I've had a process that ran for 50,000+ seconds that came up "Invalid" when none of the other results completed and - according to the site - expended 0 or close to 0 seconds which seems like they hardly got started.

So.... if the result were compared to something else that completed, I could see that but it feels like my compute cycles are being wasted here.

On World Community Grid, they compare completed results ONLY until they have a quorum. In 7,000 processes run, I haven't had one Invalid BUT I really like the idea of contributing to mapping our galaxy.

So.... can we do a better job of utilizing the contributions being made?


This could be one of the very long standing issues with Boinc, where it just stops crunching and only if you suspend the project or exit out of Boinc does it recover. OCCASIONALLY it will pick back up on its own, but the whole process is so intermittent that there are just too many things that could be causing it. It happens more often in Windows then in Linux, but it does happen in Linux too. Essentially it means you just lost that time and must now move forward to the next unit where hopefully it won't happen again. Over time we have all had this happen, I have a gpu that was crunching units normally in 22 minutes that was taking more then 24 hours, I lost tons of time. What physically is happening is that the gpu stops crunching the unit and it gets shifted to the cpu, which is MUCH slower to crunch and in some cases is just incapable of crunching the unit to it's conclusion, but it keeps trying!!


Post to thread

Message boards : Application Code Discussion : Invalid results on N-Body Simulation process


Main page · Your account · Message boards


Copyright © 2017 AstroInformatics Group