WUs run >60 hours at 0.4% // no checkpoints


Advanced search

Message boards : Problems and Help : WUs run >60 hours at 0.4% // no checkpoints

Author Message
Profile Tobias Becke [Gridcoin]
Send message
Joined: 8 Feb 15
Posts: 4
Credit: 323,097
RAC: 0
Message 13659 - Posted: 23 Mar 2015, 7:08:21 UTC

Hi,

I know my computer finished some WUs some weeks ago but I always had the same problem:

After loading a WU, the percentage rises to 0.4XX % and then stays there. The estimated time for finishing the WU constantly rises. The computer solves the WU at some point. Usually the WU was solved in a 20 hour timeframe, but now its more than 60 hours, the estimated time is 12500 hours and rising. Probably this is a bigger WU then the ones that were finished in 20 hours but I am a bit suspicious.

I have a i7 4770K @ 4.1 GHz, 16Gb RAM and WIn 7 prof.

Can I somehow change this situation of a non-displayed progress?

Another question: Is there any way to include checkpoints in the WUs? I had a Wu running for 50 hours but then had to reboot the computer. This was baaad because I had to start a new WU.

Tobi

Profile Tobias Becke [Gridcoin]
Send message
Joined: 8 Feb 15
Posts: 4
Credit: 323,097
RAC: 0
Message 13660 - Posted: 23 Mar 2015, 11:02:15 UTC

Oh I just saw, the WU exceeded the deadline. YEY :(

Profile DoctorNow
Project donor
Avatar
Send message
Joined: 11 Apr 05
Posts: 392
Credit: 2,168,338
RAC: 38
Message 13661 - Posted: 23 Mar 2015, 14:35:12 UTC - in response to Message 13659.
Last modified: 23 Mar 2015, 14:45:33 UTC

After loading a WU, the percentage rises to 0.4XX % and then stays there. The estimated time for finishing the WU constantly rises. The computer solves the WU at some point. Usually the WU was solved in a 20 hour timeframe, but now its more than 60 hours, the estimated time is 12500 hours and rising. Probably this is a bigger WU then the ones that were finished in 20 hours but I am a bit suspicious.

Hi!

Since you're relatively new to the project here are some tips:
What you experienced is usual behaviour of BURP wus. The Blender processes can be different and several steps of rendering are done in a wu, like motion blur, afterprocessing etc. The % isn't accurate for the most steps and can jump - even over 100% or get negative - so you shouldn't be too much concerned about it.
When the wu does take an unusual amount of time first check if all processors are still working on the wu. If not, there's a (slight) chance/possibility that the wu might have stalled - can happen, but does rarely.

Is there any way to include checkpoints in the WUs? I had a Wu running for 50 hours but then had to reboot the computer.

There are always sessions with wus here that can take a lot of processing time. It can be bad because the Blender app has no checkpoints (and there probably never will - depending on the developers) but it depends if you yourself can let the computer run 24/7 or not...
It's possible to save a state of a wu with the hibernate function of the pc when you really need to shutdown it, but from my experience it does not always work.

Oh I just saw, the WU exceeded the deadline. YEY :(

Don't worry. If you return the wu successfully you will get credit, regardless if it's over the deadline or not. That's a guarantee here. ;-)
____________
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats

Profile Tobias Becke [Gridcoin]
Send message
Joined: 8 Feb 15
Posts: 4
Credit: 323,097
RAC: 0
Message 13662 - Posted: 23 Mar 2015, 17:08:12 UTC - in response to Message 13661.

Thank you very much for the answer :) So 60+ hours is not unusual for 8 CPU cores? Or how can I tell whether a WU is stalled?

Greets, Tobi

Profile DoctorNow
Project donor
Avatar
Send message
Joined: 11 Apr 05
Posts: 392
Credit: 2,168,338
RAC: 38
Message 13663 - Posted: 23 Mar 2015, 18:29:33 UTC - in response to Message 13662.
Last modified: 23 Mar 2015, 18:42:06 UTC

So 60+ hours is not unusual for 8 CPU cores?

It can take that long and more, indeed. As we crunched for the movie "Big Buck Bunny" with the Sunflower-app some sessions were very computing intensive and wus could took over 100 hours.
It looks like you cancelled the wu in question earlier. I see this wu in your task list which has 66 hours.
However, it's from session 2642 and I know there are sometimes long tasks in danan's work, but after checking some of the surrounding frame wus (which didn't take that long) it could be that you might have encountered a render problem because of a cut which we had recently in another session. It caused a few tasks to render endless.
Don't know, but maybe it's the same here.

Or how can I tell whether a WU is stalled?

If you don't know how:
open the task manager (usually opens by pressing STRG, ALT and DEL together) and check the cpu usage of the blender task. If it is very low for a longer time (let's say at least 15 minutes, hard to tell sometimes) it could have been stalled.
____________
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats

Profile Tobias Becke [Gridcoin]
Send message
Joined: 8 Feb 15
Posts: 4
Credit: 323,097
RAC: 0
Message 13664 - Posted: 23 Mar 2015, 22:22:46 UTC - in response to Message 13663.

Thank you very much for the hints and greetings from Hamburg to Heidelberg :)

BTW, without making to much of advertising, it is possible to get paid for BOINC work. Have a look at Gridcoin :)


Post to thread

Message boards : Problems and Help : WUs run >60 hours at 0.4% // no checkpoints