| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: The Clean Energy Project - Phase 2 Forum
Thread: Poor crunching on this project |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 156
|
|
| Author |
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
On the swapfile use, so far found that BOINC is doing rather nothing in practise with the percent. Certainly not in 6.10.58 nor in the alpha 6.12.2 client (strictly for testing). Early in starting on Linux I did bookmark this little manual on swapfiles https://help.ubuntu.com/community/SwapFaq and have since introduced the swappiness value into the sysctl.conf file and shortly be doing a swapoff if nothing betters, and to complete the tests, move swapfile into RAM. I just want those 30-60 minutes per task back... simple as that.
----------------------------------------have a nice day. PS: Have put a query in with the alpha mail list to ascertain what if any the swap space setting still has any functionality. Will share if anything new is learned.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
yose-ue
Cruncher Joined: Dec 27, 2008 Post Count: 21 Status: Offline Project Badges: 
|
The linux kernel is effecting the efficiency when running CEP2 jobs at least on my system.
I have a choice of two linux kernels when I boot my system. when I boot with linux kernel 2.6.32-25-generic I have only 26% efficiency when crunching CEP2 jobs. when I boot with linux kernel 2.6.32-24-generic I have about 94% efficiency. with the older kernel at least it is practical to run CEP2 jobs it is not as efficient as most other types of jobs which are over 99% efficient (when not using the computer for other things). I couldn't justify loosing nearly three quarters of my time just to run this project. |
||
|
|
I need a bath
Senior Cruncher USA Joined: Apr 12, 2007 Post Count: 347 Status: Offline Project Badges: 
|
Hmm I have a choice between 2.6.35-22
----------------------------------------and 2.6.32-25 ANy idea how to get back the older kernel without reinstalling the entire OS? nevermind. I did it and still FUBAR with 2.6.32-24   [Edit 1 times, last edit by I need a bath at Oct 21, 2010 9:28:40 PM] |
||
|
|
RaymondFO
Veteran Cruncher USA Joined: Nov 30, 2004 Post Count: 561 Status: Offline Project Badges:
|
Hmm I have a choice between 2.6.35-22 and 2.6.32-25 ANy idea how to get back the older kernel without reinstalling the entire OS? nevermind. I did it and still FUBAR with 2.6.32-24 Did you do a clean install or attempt a downgrade in the kernel? If not a clean install, I have a new experiment: A clean April 2010 OS install with NO updates. I have not tried that approach. |
||
|
|
I need a bath
Senior Cruncher USA Joined: Apr 12, 2007 Post Count: 347 Status: Offline Project Badges:
|
not clean install. I'm not going to do that. I can only take so much!
---------------------------------------- |
||
|
|
RaymondFO
Veteran Cruncher USA Joined: Nov 30, 2004 Post Count: 561 Status: Offline Project Badges:
|
I will do that as well. Very curious to see what happens with this experiment. The old Linux OS download has kernel 2.6.32.21-generic. My other Linux computer has 2.6.35-22 generic. Not upgrading OS on install.
----------------------------------------Let's see how these CEP2 WU's perform: E200479_ 679_ A.24.C19H12N2S2Se.96.0.set1d06_ 0-- Catseye In Progress 10/22/10 02:54:17 11/1/10 02:54:17 0.00 0.0 / 0.0 E200479_ 626_ A.24.C19H12N2S2Se.355.4.set1d06_ 1-- Catseye In Progress 10/22/10 02:53:56 11/1/10 02:53:56 0.00 0.0 / 0.0 E200479_ 671_ A.24.C19H12N2S2Se.22.0.set1d06_ 0-- Catseye In Progress 10/22/10 02:53:56 11/1/10 02:53:56 0.00 0.0 / 0.0 E200479_ 576_ A.24.C19H12N2S2Se.220.0.set1d06_ 0-- Catseye In Progress 10/22/10 02:39:09 11/1/10 02:39:09 0.00 0.0 / 0.0 [Edit 2 times, last edit by RaymondFO at Oct 22, 2010 2:56:28 AM] |
||
|
|
RaymondFO
Veteran Cruncher USA Joined: Nov 30, 2004 Post Count: 561 Status: Offline Project Badges:
|
With wall clock times running times around 6:24:00, the results are definitely an improvement, and the best I have seen since this issue started to occur:
E200479_ 576_ A.24.C19H12N2S2Se.220.0.set1d06_ 0-- Catseye Pending Validation 10/22/10 02:39:09 10/22/10 09:55:15 4.97 99.8 / 0.0 E200479_ 671_ A.24.C19H12N2S2Se.22.0.set1d06_ 0-- Catseye Pending Validation 10/22/10 02:53:56 10/22/10 09:55:15 4.94 99.3 / 0.0 E200479_ 679_ A.24.C19H12N2S2Se.96.0.set1d06_ 0-- Catseye Pending Validation 10/22/10 02:54:17 10/22/10 09:55:15 5.52 110.8 / 0.0 E200479_ 626_ A.24.C19H12N2S2Se.355.4.set1d06_ 1-- Catseye Valid 10/22/10 02:53:56 10/22/10 09:55:15 5.07 101.9 / 142.5 Maybe it is the kernel. Will continue to run CEP2 with other WU projects to see if further improvements can occur. |
||
|
|
genes
Advanced Cruncher USA Joined: Jan 28, 2006 Post Count: 132 Status: Offline Project Badges: 
|
Here's my take on the "lost time" issue. Firstly, I'm running Ubuntu 10.04 32-bit on a C2D 6700, 2.66GHz. 4GB memory installed, 3.2GB available. 2 tasks running, only CEP2, 24/7. Yesterday I took out a notepad (you know, made of paper) and followed a particular task until it finished, periodically checking the "properties" and writing down elapsed time, CPU time, and %progress. Here are the results:
----------------------------------------elapsed_____CPU_______%progress 1:06:18_____1:04:12_____8.917 1:15:02_____1:12:49____10.114 1:30:26_____1:27:57____12.218 2:00:22_____1:57:23____16.304 ----- then something happened while I was having dinner. 2:56:37_____0:39:11_____5.444 3:00:05_____0:42:35_____5.916 3:30:04_____1:05:01_____9.031 ----- another checkpoint! "CPU time at last checkpoint - 00:50:34" 3:38:03_____0:50:49_____7.058 3:53:18_____1:05:49_____9.143 4:00:40_____1:13:02____10.146 4:30:40_____1:42:33____14.244 4:45:28_____1:57:07____16.267 ----- another checkpoint! "CPU time at last checkpoint - 00:54:43" 4:46:07_____0:54:58_____7.635 5:10:15_____1:18:44____10.936 5:32:03_____1:40:10____13.914 ----- another checkpoint! "CPU time at last checkpoint - 00:58:51" 5:53:55_____1:06:47_____9.277 6:00:05_____1:12:52____10.120 6:35:06_____1:47:22____14.913 6:58:24_____2:10:06 ----- Task finished. I checked with the error log, there were no repeated subtasks, just normal processing. This was from task E200478_771_A.24... (don't know if others can see this, give it a shot) (edit) Anyway, the time resets that occur don't show up in the log other than that "the CPU time was restored to xx". Why are they doing this? Only 28% of the actual CPU time spent (not elapsed time!) was reported and credited. [Edit 1 times, last edit by genes at Oct 22, 2010 11:55:37 AM] |
||
|
|
RaymondFO
Veteran Cruncher USA Joined: Nov 30, 2004 Post Count: 561 Status: Offline Project Badges:
|
Genes,
What kernel version are you using? What you have stated above is consist with those who are experiencing this CEP2 WU issue. |
||
|
|
genes
Advanced Cruncher USA Joined: Jan 28, 2006 Post Count: 132 Status: Offline Project Badges:
|
I'm using the kernel that gets installed by 10.04, which is 2.6.32.25.
|
||
|
|
|