1) Message boards : Number crunching : Long running work units (Message 1166)
Posted 8 May 2016 by fractal
Post:
This is what BM reports, but I not really believe in the numbers... Just see CPU usage...


CPU usage is normal, as in 98% or higher. I see no problems in the longer running WU's, only that they run longer... excepting the ones that never finish! Hope you can figure those out.

8-)

Yeah, the never finishing ones are a pain in the rear.

Sad to say, but I abort anything with "0" or "10" in the third field on sight. Some of them do finish but some of them don't, and it is hard to tell which is which. Turning off "keep suspended in memory", suspending and resuming them resets them back to the last check point which is often days earlier. So, rather than wait for a day to find out, I just abort any "0" or "10" sequence work and hope someone with an AMD processor picks them up.
2) Message boards : Number crunching : Long running work units (Message 1057)
Posted 30 Jan 2016 by fractal
Post:
I have observed these vastly longer running work units too. But you can see (notably in your examples) they have one common ground:
universe_bh_AAA_10_CCCCC_D-EEEEEE_FFFFFF

All these work units carry the number 10 in position B! No matter what value the other positions have.
Look in your work unit lists to verify this.

Good observation. Not sure what that field means but, yes, all of the ones I spotted seem to have a "10" there. I wonder what it is about those work units that cause some machines to take so much longer than others.

edit: I just went through 1000 of the 8000 units in my "valid tasks" list and you are 33 for 33. All 33/1000 have 10 in that position. http://universeathome.pl/universe/workunit.php?wuid=3676163 is interesting in it taking 2x as long as it should. Maybe 1/3 of the 33 took 2x as long as normal and the rest taking 8x as long. Occasionally the wingman takes the normal time, sometimes it takes 2x as long and rarely both of us take the full extra long time.

Oh, and every unit gets 333 credits no matter how long they took to compute.
3) Message boards : Number crunching : Long running work units (Message 1053)
Posted 30 Jan 2016 by fractal
Post:
Am I the only one getting the long running work units mixed in with regular running units? They appear to come in bunches.

The 333 credit work units usually take between 6000 and 12000 seconds on the various machines I have but occasionally I get the 60,000 plus second work unit. These get the same 333 credits when the complete.

Examples include http://universeathome.pl/universe/workunit.php?wuid=3718041, http://universeathome.pl/universe/workunit.php?wuid=3717416, http://universeathome.pl/universe/workunit.php?wuid=3717415http://universeathome.pl/universe/workunit.php?wuid=3717828 on a stock i7. Ok, let's ignore that last one. That's an ARM processor which took as long as the i7 ...

A stock i3 sees http://universeathome.pl/universe/workunit.php?wuid=3675859 and http://universeathome.pl/universe/workunit.php?wuid=3675872

As you can see, sometimes the wingmen also get the really long run times, sometimes they get slightly longer runtimes and sometimes they get standard runtimes.

All my machines run Intel processors at stock frequencies with decent cooling under Linux. There is no pattern to the wingmen as they run a variety of processors and operating systems.

I have written it off as "one of those things" for a while until I looked into it this weekend and found that the wingmen are not affected the same way I am.

So, does anyone else see a pile of these day long work units mixed in with their 2-3 hour units? Does anyone have any idea what is going on?







Copyright © 2021 Copernicus Astronomical Centre of the Polish Academy of Sciences
Project server and website managed by Krzysztof 'krzyszp' Piszczek