| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: decreasing average performance over time? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 9
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi,
I have a workstation that runs 24/7, and if I look at the statistics page in BOINC, choosing the host average, I see that the performances have a clear downward tendency over the last 30 days period, of about 10%. In the past, there were deviations from an average performance, due to PVs and stuff, but now the trend is clear and keeps going down. Questions: can this be related to the relative "aging" of the machine vs. the newer CPUs available to the grid, i.e., the performances of the machine are unchanged but when compared to performances of the other machines reporting results this machine is now becoming a low(er) performer and thus the reduction in average credits? Or is it something related to the machine itself? I.e., CPUs are performing worse with time, or it could be an issue related to dust on the fans? Note that the machine does not use TThrottle or similar, and the load appears to be 100% at all times. Sorry not to be very clear in describing this. |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7668 Status: Offline Project Badges: 
|
There are several things which come to mind. First, over time you may have had some additional background processes which are using processor cycles. You can check task manager or other task related programs to see what is running and what is using the cycles. Second, you may have some deterioration of of some of your memory. It would not hurt to check that with memtest. Third, the changing nature of the work units may have had a slight negative effect the credits given. Fourth, it could be heat related - dust buildup on fans, heat sinks, vents, etc. Fifth, if you have had any significant power fluctuations there may have been a weakening of some miscellaneous component of enough magnitude to affect internal communication channels in your machine.Sixth, your hard drive may be getting crowded, cluttered and/or extremely fragmented. Hope this helps. i would be interested if you do find the cause.
----------------------------------------Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
HI,
thanks for the quick reply. I checked #6 and #1, and does not seem the case. #3 and #4 could be. Now I restarted, it was a while since last time. #1 or #5 are a very possible reasons, as the machine is 5 years old, and last 2 1/2 has been running 24/7. I will look into it. Thanks again |
||
|
|
yojimbo197
Advanced Cruncher Joined: Jun 30, 2012 Post Count: 83 Status: Offline Project Badges: 
|
The most noticeable thing I've seen is a 2-4 minute decrease in HCC CPU_GPU jobs after cleaning out my GPU fan and heatsink. Supposedly defragmenting your HD doesn't significantly speed up WCG jobs if you believe previous posts from WCG members.
----------------------------------------I'd be curious to know if anyone has seen a repeated and significant decrease in WCG job completion times when the CPU core temps at 100% load are decreased.  |
||
|
|
Bearcat
Master Cruncher USA Joined: Jan 6, 2007 Post Count: 2803 Status: Offline Project Badges: 
|
Can you post the specs or your workstation with OS used?
----------------------------------------
Crunching for humanity since 2007!
 |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The most noticeable thing I've seen is a 2-4 minute decrease in HCC CPU_GPU jobs after cleaning out my GPU fan and heatsink. Supposedly defragmenting your HD doesn't significantly speed up WCG jobs if you believe previous posts from WCG members. I'd be curious to know if anyone has seen a repeated and significant decrease in WCG job completion times when the CPU core temps at 100% load are decreased. What is the consequence of "too hot" on e.g. power leakage and the protection mechanism of many modern CPU's? Certainly, on my laptop I see the GHZ being decremented in steps of about 100MHz, till it's not too hot anymore. So, too hot can impact production/throughput. Anyone not following the dustbunnies fighter advise [best, every 3 months even if not too dusty an environment] is in for reduced live-span and less work completed. As for HD defragmenting... depends on the size of the job and on the amount of IO a task does and the general state of the drive and it's defrag maintenance [under Windows]. Best location for BOINC data_dir is it's own drive, but at least it's own logical partition. That will work to kind of self regulate fragmentation under Windows [mostly prevent it]. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello latakia
Reference: latakia [Dec 9, 2012 8:25:49 PM] post Any hardware's performance is bound to degrade over time from use. Valuation of WUs is mostly heavy on the hardware side of things. The software side, while broadly minimal, does play a role in the valuation. The wild card involves the valuation process itself and that covers: the tolerances, adjustments vis-a-vis 'equilibrate', the calibration of reference values, and the 'distortion' introduced by the quorum system. If I have to guess, 10%-20% deviation would be just about 'right' to account for the 'fog-of-valuation' that is not traceable to hardware or software sources. ; ; andzgridPost#728 ; |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi Bearcat,
you will like the kind of CPU, but it's an old one ;-) Processor: 4 GenuineIntel Intel(R) Xeon(R) CPU E5440 @ 2.83GHz [Family 6 Model 23 Stepping 6] Processor: 6.00 MB cache Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 nx lm vmx tm2 dca pbe OS: Microsoft Windows XP: Professional x86 Edition, Service Pack 3, (05.01.2600.00) |
||
|
|
Bearcat
Master Cruncher USA Joined: Jan 6, 2007 Post Count: 2803 Status: Offline Project Badges:
|
Ah, harpertowns. Had 3 rigs with those in them. Those FSB dimms run extremely hot. Put heat spreaders on and ram fans to help cool them. Bet this is your problem. Also, how are your cpu temps? Would suggest to re apply thermal paste too.
----------------------------------------
Crunching for humanity since 2007!
|
||
|
|
|