| 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: CEP2 Quick CPU Processing Times? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 33
|
|
| Author |
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
hmmm, what BOINC is this from on what *nix distro? Since the CEP2 tasks relaunch for each job within a task under Linux, I wonder if your particular setup misses to catch the time, not always, when a new loop starts. Would not know where to point finger, Science App or BOINC, likely not OS. So, I've run so far with the proper distro on Lucid Lynx, first 6.10.17 and now 6.10.58 (Debian package from the getdeb ppa). They ran/run fine and honky dory. The
----------------------------------------edit: I hate English homophones
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Jul 30, 2010 2:29:30 PM] |
||
|
|
codes
Advanced Cruncher Joined: Oct 20, 2009 Post Count: 142 Status: Offline |
what BOINC is this from on what *nix distro? The BOINC is version 6.10.17 from the WCG site running on Slackware 13.1 x86. I'm in the process of installing the BOINC WCG client now, will report back after the first CEP2 wu get's processed. |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3716 Status: Offline Project Badges: 
|
Other than CEP2 wu's, all other WCG application wu's result in equal CPU and Elapsed time completed for the respective wu processed. So I suspect the problem is software related to CEP2 on my computer. OK, this is eliminating a general CPU time reporting problem in your machine.And since CPU times are normal for the other applications Sekerob's idea of a problem when chaining the jobs inside a CEP2 WU is worth an investigation by the techs. |
||
|
|
codes
Advanced Cruncher Joined: Oct 20, 2009 Post Count: 142 Status: Offline |
Sekerob's idea of a problem when chaining the jobs inside a CEP2 WU is worth an investigation by the techs Is there a user account privilege requirement for the running application and/or working directory location placement requirement for CEP2? How about write permissions and specific location of a scratch temporary directory, such as /tmp? Mine is running under a non-root user account and located in the user account home directory. Location: /home/codes/BOINC All directories/files under /home/codes are owned and writable by account "codes". The computer temporary /tmp directory has the standard sticky permissions set on it, therefore non-root account "codes" does have the capabilty to write to it. I have checked all the system logs on the computer and nothing found relating to any type of BOINC application/process error. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Cruncher
I am running 6.10.52 on an install of Ubuntu 10.10-testing during the day and what ever the repo version (not on there right now so can't check) on Ubuntu 9.04 at night. I am running at 98% on 9.04 and 95% on 10.10. My times seem to be about the same between the two install and they seem similar to clock time. Both these installs are on the same box. 9.04 is on my main drive that I keep turned off in bios when running testing on an external, rough stats in sig on box. The install on 10.10 is just like yours, in my home directory. I think this is a much better set up than what Ubuntu does with the install from the package manager. It is also portable which is nice if I screw this install, I have a back up with the same user name and can switch and nothing even indicates a switch in the OS. I like that as there is not another device going, or another OS, just another install of the same OS on the same drive. I think, by the way, that under 9.04 I am using the same version you are but with it scattered all over the place by the Ubuntu installer. You might want to try a little newer version and see if that makes a difference. Seems awfully fast to me. My times are usually a little better than the wing man but not always. Have credit for 72. Believe that I have one error and it sounds like the one you had. Can't check that now as they are updating. |
||
|
|
codes
Advanced Cruncher Joined: Oct 20, 2009 Post Count: 142 Status: Offline |
E200181_ 810_ A.19.C16H11NSeSi.83.2.set1d06_ 2-- 619 Valid 7/29/10 16:00:00 7/29/10 23:22:13 1.20 <==== Me
E200181_ 810_ A.19.C16H11NSeSi.83.2.set1d06_ 0-- 619 Valid 7/19/10 15:28:30 7/20/10 08:00:33 3.88 Resetting the BIOS, reinstalling the OS, and BOINC Manager 6.10.17 made no difference. The CPU time is still reporting low, 1.2 Hours. According to the "Properties" of the processed WU, the Elapsed time was 4 Hours and 51 Minutes. I removed BOINC Manager 6.10.17 and installed 6.10.56 Manager from the Berkeley site to see if that makes a difference. I also installed the same Slackware Linux 13.1 x86 OS on another computer I have. It's a slower one, GenuineIntel Intel(R) Celeron(R) CPU 430 @ 1.80GHz [Family 6 Model 22 Stepping 1], but will see if the CPU and Elapsed times are closer together than what I'm seeing with the 2.6 GHz dual processor I seem to be having problems with. |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3716 Status: Offline Project Badges:
|
codes,
----------------------------------------All files used by a task are in the corresponding slot subdirectory in the "slots" subdirectory in the BOINC Data directory and belong to the user running the BOINC client. I am running 6.10.17 and have never seen any similar problem with CEP2 in this machine, so I think this BOINC version can be trusted, at least under Ubuntu 10.04. We'll see what's the result of your installing the same OS in another machine. Edit: I have posted a heads-up to the techs in the CA forum. ---------------------------------------- [Edit 1 times, last edit by JmBoullier at Jul 30, 2010 3:19:49 AM] |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
codes,
We are looking into this situation. It is as if your computer CPU time is not being recorded properly between check points. I seriously doubt that your machine is taking 3 hours of IO time that is not being recorded. Unfortunately we do not have a direct answer for you at the moment. We will try to get you an answer as soon as we know something. Thanks, -Uplinger |
||
|
|
codes
Advanced Cruncher Joined: Oct 20, 2009 Post Count: 142 Status: Offline |
Still reported lower CPU on a different Boinc Manager and computer.
----------------------------------------Boinc Manager 6.10.56. Properties of computed WU show CPU time 1 Hr 17 Min and Elapased time 4 Hr 42 Min: E200215_ 315_ A.21.C19H12N2.207.0.set1d06_ 1-- 619 Valid 7/30/10 22:30:30 7/31/10 12:17:19 1.29 <== Me - Pentium Dual 2.6 GHz E200215_ 315_ A.21.C19H12N2.207.0.set1d06_ 0-- 619 Valid 7/30/10 22:25:37 7/31/10 05:14:06 6.19 Boinc Manager 6.10.17. Properties of computed WU show CPU time 2 Hr 7 Min and Elapased time 8 Hr 40 Min: E200214_ 746_ A.21.C18H15NSSi.3.3.set1d06_ 0-- 619 Valid 7/30/10 16:49:01 7/31/10 12:13:42 2.13 <== Me - Celeron 1.8 GHz E200214_ 746_ A.21.C18H15NSSi.3.3.set1d06_ 1-- 619 Valid 7/30/10 16:48:45 7/31/10 11:05:54 3.94 I installed the same Slackware Linux 13.1 x86 on the Celeron. Both have 80 GB sata drive EXT4 partitioned 2 GB swap, / 40 GB, and /home 38 GB. On both I exit out of the GUI when done looking at the Boinc Manager, so the majority of time all Boinc processes are running in CLI environment. Both are using the "Default" device profile with "Maximum Output" set. Both have "Use no more than: 15% of total disk space" (You may recall previously this was set to the default 10%, I changed it to 15% to start getting CEP2 wu's when /home partition was set to 20 GB). Both are dedicated 24/7 to WCG. Both show in "top" command 99% CPU usage for all Boinc related processes. Maybe the problem is related to Slackware Linux 13.1 x86. Seems kinda strange as Slackware is the most generic Linux distro. So now I'm removing Slackware from both these computers and putting OpenSuSE 11.3 x86 on them. Let's see if anything changes... [Edit 1 times, last edit by codes at Jul 31, 2010 1:21:55 PM] |
||
|
|
codes
Advanced Cruncher Joined: Oct 20, 2009 Post Count: 142 Status: Offline |
Still under reporting CPU time. I installed Fedora 13 x86 on the Celeron 1.8 GHz computer, this is my second computer. The Pentium 2.6 GHz that I originally reported the problem with has Windows on it now crunching non-CEP2 WU's.
Boinc Manager 6.10.45. Properties of computed WU show CPU time 3 Hr 2 Min and Elapased time 10 Hr 37 Min: E200217_ 543_ A.22.C14H6N4S4.13.2.set1d06_ 1-- 619 Valid 7/31/10 18:45:53 8/1/10 13:30:20 3.04 <== Me E200217_ 543_ A.22.C14H6N4S4.13.2.set1d06_ 0-- 619 Valid 7/31/10 18:45:17 8/1/10 06:11:01 7.21 <core_client_version>6.10.45</core_client_version> <![CDATA[ <stderr_txt> INFO: No state to restore. Start from the beginning. [14:50:31] Number of jobs = 16 [14:50:31] Starting job 0,CPU time has been restored to 0.000000. [14:50:31] Starting new Job [14:50:31] Qink name = fldman [14:50:31] Qink name = gesman [14:50:31] Qink name = scfman Quit requested: Exiting INFO: No state to restore. Start from the beginning. [15:01:04] Number of jobs = 16 [15:01:04] Starting job 0,CPU time has been restored to 0.000000. [15:01:22] Starting new Job [15:01:24] Qink name = fldman [15:01:25] Qink name = gesman [15:01:25] Qink name = scfman [15:04:49] Qink name = anlman [15:04:50] End of Job [15:04:53] Finished Job #0 [15:04:53] Starting job 1,CPU time has been restored to 61.219693. [15:04:53] Starting new Job [15:04:53] Qink name = fldman [15:04:54] Qink name = gesman [15:04:54] Qink name = scfman [15:14:51] Qink name = anlman [15:16:20] End of Job [15:16:23] Finished Job #1 [15:16:23] Starting job 2,CPU time has been restored to 124.530068. [15:16:24] Starting new Job [15:16:24] Qink name = fldman [15:16:24] Qink name = gesman [15:16:24] Qink name = scfman [15:26:30] Qink name = anlman [15:26:30] Qink name = drvman [15:29:10] Qink name = optman [15:29:10] Qink name = fldman [15:29:10] Qink name = gesman [15:29:11] Qink name = scfman [15:40:45] Qink name = anlman [15:40:45] Qink name = drvman [15:42:45] Qink name = optman [15:42:45] Qink name = fldman [15:42:45] Qink name = gesman [15:42:45] Qink name = scfman [16:20:12] Qink name = anlman [16:20:12] Qink name = drvman [16:33:02] Qink name = optman [16:33:02] Qink name = fldman [16:33:02] Qink name = gesman [16:33:03] Qink name = scfman [16:43:06] Qink name = anlman [16:43:06] Qink name = drvman [16:45:02] Qink name = optman [16:45:02] Qink name = fldman [16:45:02] Qink name = gesman [16:45:02] Qink name = scfman [16:54:06] Qink name = anlman [16:54:06] Qink name = drvman [16:56:02] Qink name = optman [16:56:02] Qink name = fldman [16:56:02] Qink name = gesman [16:56:03] Qink name = scfman [17:03:30] Qink name = anlman [17:03:30] Qink name = drvman [17:05:26] Qink name = optman [17:05:26] Qink name = fldman [17:05:26] Qink name = gesman [17:05:26] Qink name = scfman [17:12:10] Qink name = anlman [17:12:10] Qink name = drvman [17:14:05] Qink name = optman [17:14:05] Qink name = anlman [17:14:26] End of Job [17:14:28] Finished Job #2 [17:14:28] Starting job 3,CPU time has been restored to 188.833292. [17:14:28] Starting new Job [17:14:28] Qink name = fldman [17:14:29] Qink name = gesman [17:14:29] Qink name = scfman [17:23:03] Qink name = anlman [17:23:22] End of Job [17:23:25] Finished Job #3 [17:23:25] Starting job 4,CPU time has been restored to 250.918853. [17:23:25] Starting new Job [17:23:25] Qink name = fldman [17:23:26] Qink name = gesman [17:23:26] Qink name = scfman [17:29:37] Qink name = anlman [17:29:57] End of Job [17:30:00] Finished Job #4 [17:30:00] Starting job 5,CPU time has been restored to 315.206079. [17:30:00] Starting new Job [17:30:00] Qink name = fldman [17:30:00] Qink name = gesman [17:30:00] Qink name = scfman [17:36:32] Qink name = anlman [17:36:51] End of Job [17:36:54] Finished Job #5 [17:36:54] Starting job 6,CPU time has been restored to 379.090367. [17:36:54] Starting new Job [17:36:54] Qink name = fldman [17:36:55] Qink name = gesman [17:36:55] Qink name = scfman [17:42:53] Qink name = anlman [17:43:13] End of Job [17:43:15] Finished Job #6 [17:43:15] Starting job 7,CPU time has been restored to 442.846674. [17:43:15] Starting new Job [17:43:15] Qink name = fldman [17:43:16] Qink name = gesman [17:43:16] Qink name = scfman [17:52:29] Qink name = anlman [17:52:49] End of Job [17:52:52] Finished Job #7 [17:52:52] Starting job 8,CPU time has been restored to 506.969925. [17:52:52] Starting new Job [17:52:52] Qink name = fldman [17:52:53] Qink name = gesman [17:52:53] Qink name = scfman [17:58:03] Qink name = anlman [17:58:22] End of Job [17:58:25] Finished Job #8 [17:58:25] Starting job 9,CPU time has been restored to 571.315143. [17:58:25] Starting new Job [17:58:26] Qink name = fldman [17:58:26] Qink name = gesman [17:58:26] Qink name = scfman [18:05:51] Qink name = anlman [18:06:25] End of Job [18:06:28] Finished Job #9 [18:06:28] Starting job 10,CPU time has been restored to 635.214428. [18:06:28] Starting new Job [18:06:28] Qink name = fldman [18:06:28] Qink name = gesman [18:06:29] Qink name = scfman [18:21:52] Qink name = anlman [18:22:26] End of Job [18:22:29] Finished Job #10 [18:22:29] Starting job 11,CPU time has been restored to 698.980734. [18:22:29] Starting new Job [18:22:29] Qink name = fldman [18:22:30] Qink name = gesman [18:22:30] Qink name = scfman [18:30:00] Qink name = anlman [18:30:34] End of Job [18:30:37] Finished Job #11 [18:30:37] Starting job 12,CPU time has been restored to 762.419089. [18:30:37] Starting new Job [18:30:37] Qink name = fldman [18:30:40] Qink name = gesman [18:30:41] Qink name = scfman [19:20:13] Qink name = anlman [19:24:09] End of Job [19:24:13] Finished Job #12 [19:24:13] Starting job 13,CPU time has been restored to 825.861444. [19:24:13] Starting new Job [19:24:13] Qink name = fldman [19:24:16] Qink name = gesman [19:24:16] Qink name = scfman [21:28:02] Qink name = anlman [21:31:58] End of Job [21:32:01] Finished Job #13 [21:32:01] Starting job 14,CPU time has been restored to 888.334946. [21:32:01] Starting new Job [21:32:01] Qink name = fldman [21:32:04] Qink name = gesman [21:32:05] Qink name = scfman Quit requested: Exiting [21:37:35] Number of jobs = 16 [21:37:35] Starting job 14,CPU time has been restored to 888.334946. [21:37:57] Starting new Job [21:37:58] Qink name = fldman [21:38:01] Qink name = gesman [21:38:02] Qink name = scfman [23:29:04] Qink name = anlman [23:33:09] End of Job [23:33:11] Finished Job #14 [23:33:11] Starting job 15,CPU time has been restored to 950.766454. [23:33:12] Starting new Job [23:33:12] Qink name = fldman [23:33:15] Qink name = gesman [23:33:15] Qink name = scfman [02:14:54] Qink name = anlman [02:21:47] End of Job [02:21:50] Finished Job #15 called boinc_finish Exiting 0 </stderr_txt> ]]> Dosen't make sense, both of my computers crunch different app WU's correctly and report CPU time accurately except CEP2 under Linux. Even with a different Linux distro the same inaccurate CPU time behavior appears while other CEP2 crunchers don't seem to have a problem. I installed the "boinc-client" using the Fedora Software Manager: working directory /var/lib/boinc-client, user account "boinc". I turned off the screensaver.  I'll give Ubuntu a try, but need to get some more blank DVD's so I can burn the iso. |
||
|
|
|