| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 11
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello all
I recently was introduced to the project by a friend, and on doing so put into service some older quad core quad socket IBM servers. (x3850m2) When i first started with only 1 machine running, I was getting work unit times of around 5-6 hours. Installed/setup second server, times increased to 10 hours per unit As above third server, times are now increasing to around 20 hours per unit Am I being throttled? or has there been a genuine increase in work unit size last few days? it almost seems like as soon as I increased my horsepower the times changed? is it just me? new, so trying to get an idea on how things work :) Regards, Alex |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi, probably just you... no you're not being throttled, by WCG
![]() Nothing indicates a 3 fold plus runtime increase [Chart]. Of course the different sciences have different and varying runtimes. Maybe if you post a copy of the Result Status page results, we can have a look at Elapsed [Runtime] versus CPU time ratios... the efficiency at which jobs run. Are you running a random mix of researches and does that include Clean Energy? Latter is hard work, the more are run concurrent [on same host], the poorer the efficiency, [bar some manual optimizing of configs]. TTYL |
||
|
|
Warped@RSA
Senior Cruncher South Africa Joined: Jan 15, 2006 Post Count: 440 Status: Offline Project Badges:
|
I have an old laptop which runs MCM exclusively.
----------------------------------------The work units have once again increased in length quite dramatically. See the the thread entitled "Much Longer WUs" for a discussion on the subject.
Dave
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
So indeed, after almost 2 months of near flatlining at about 5.4 hours, it rocketed to a mean of 7.28 for the day, a new project record. Looking at the laptop history they took 20 hours on the 26th, 14-18 hours on the 27th, 4-8 hours on the 28th [batches 13346-51 repairs]. A task of 13499 is aiming for 20 hours again.
![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
thanks for your advice people, the units in question are saying they are MCM ones, needing aprox 48k GFLOPS, the time difference between 3 identical bits of hardware is a bit funny though?
maybe the machine running windows OS instead of linux is at a disadvantage somehow? i'll screenie my results page and post for all to see. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Results Clip, As promised
![]() Like i was saying, the 3 machines are mostly identical hardware if anything spaceball3 (which is running W2008 server) is the highest spec, with xeon E7440s, the other two machines are E7330s all 2.4ghz quad cores with 4 sockets each, total 16 cores per machine, 64GB ram Regards, Alex |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
In addition, i've noticed that all 3 machines are starting to pick up WUs that are estimated 18 hours plus.
Regards again, Alex |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The short form conclusion is: Coincidence, so happened you were gearing up as the MCM got longer.
You have an OET and FAHV inbetween the first 15 taking which show excellent runtimes. My Linux takes about 30 to 45 minutes for latter... 0.5 to 0.75 hours, Q6600 from 2005 or so. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks rob, i guess it was just a massive co-incidence :)
|
||
|
|
Seoulpowergrid
Veteran Cruncher Joined: Apr 12, 2013 Post Count: 823 Status: Offline Project Badges:
|
I must admit I love the Spaceball names for your machines :)
----------------------------------------![]() |
||
|
|
|