Comments for Freeminded.org http://www.freeminded.org ... think - be free! Wed, 16 May 2012 08:22:54 +0000 hourly 1 http://wordpress.org/?v=3.3.2 Comment on 5.11 Sabre 2.0 – numerous improvements! by 5.11 Zipper-pull tabs: sell them separately please @ Freeminded.org http://www.freeminded.org/index.php/2012/04/5-11-sabre-2-0-numerous-improvements/comment-page-1/#comment-1192 5.11 Zipper-pull tabs: sell them separately please @ Freeminded.org Wed, 16 May 2012 08:22:54 +0000 http://www.freeminded.org/?p=881#comment-1192 [...] I acquired a 5.11 Sabre 2.0 Jacket which sports some really nice zipper-pull tabs. And since 5.11 is known for listening to its [...] [...] I acquired a 5.11 Sabre 2.0 Jacket which sports some really nice zipper-pull tabs. And since 5.11 is known for listening to its [...]

]]> Comment on phpmyadmin on (K)Ubuntu 10.10 not working? by Skidder http://www.freeminded.org/index.php/2011/02/phpmyadmin-on-kubuntu-10-10-not-working/comment-page-1/#comment-1191 Skidder Wed, 16 May 2012 07:40:10 +0000 http://www.freeminded.org/?p=332#comment-1191 Great! thanks for the reply. Great! thanks for the reply.

]]> Comment on phpmyadmin on (K)Ubuntu 10.10 not working? by Dinesh http://www.freeminded.org/index.php/2011/02/phpmyadmin-on-kubuntu-10-10-not-working/comment-page-1/#comment-1189 Dinesh Wed, 16 May 2012 02:21:47 +0000 http://www.freeminded.org/?p=332#comment-1189 If you are facing some problem with phpmyadmin then consult this post [http://codesprints.blogspot.in/2012/04/phpmyadmin-not-working-ubuntu.html]. If you are facing some problem with phpmyadmin then consult this post [http://codesprints.blogspot.in/2012/04/phpmyadmin-not-working-ubuntu.html].

]]> Comment on OMNeT++ and how to use it with a database by Clustercomputing with Torque @ Freeminded.org http://www.freeminded.org/index.php/2011/02/omnet-and-how-to-use-it-with-a-database/comment-page-1/#comment-1167 Clustercomputing with Torque @ Freeminded.org Mon, 07 May 2012 08:42:47 +0000 http://www.freeminded.org/?p=327#comment-1167 [...] Fourth: Get your results. You may either opt for log files (.sca, .elog, .vec etc. are written to the networked filesystem) or have a look at how to output OMNeT++ results to a database. [...] [...] Fourth: Get your results. You may either opt for log files (.sca, .elog, .vec etc. are written to the networked filesystem) or have a look at how to output OMNeT++ results to a database. [...]

]]> Comment on Clustercomputing with Torque by Skidder http://www.freeminded.org/index.php/2011/05/clustercomputing-with-torque-building-a-job-submission-file/comment-page-1/#comment-1157 Skidder Fri, 04 May 2012 09:10:14 +0000 http://www.freeminded.org/?p=525#comment-1157 In the PBS you can specify you have a job which requires a certain number of CPUs. The opp_runall -j60 basically spawns 60 <em>independent</em> processes, so if one fails the other continue. These processes are distributed among the number of CPUs (note that the -j should not be larger than the number of CPUs specified in the PBS or you'll get a lot of context switches which is detrimental to performance). An alternative is, as you mention, to create a number of PBS files. You could create one for each simulation run (then you do not need opp_runall). This may give the Torque/MAUI batch scheduler a finer granularity to work with: If I claim 60 CPUs for 60 processes, and there are 10 which take considerably longer than the other 50, I still have claimed these CPUs. That's something to keep in mind on busy clusters. In the PBS you can specify you have a job which requires a certain number of CPUs. The opp_runall -j60 basically spawns 60 independent processes, so if one fails the other continue. These processes are distributed among the number of CPUs (note that the -j should not be larger than the number of CPUs specified in the PBS or you’ll get a lot of context switches which is detrimental to performance).

An alternative is, as you mention, to create a number of PBS files. You could create one for each simulation run (then you do not need opp_runall). This may give the Torque/MAUI batch scheduler a finer granularity to work with: If I claim 60 CPUs for 60 processes, and there are 10 which take considerably longer than the other 50, I still have claimed these CPUs. That’s something to keep in mind on busy clusters.

]]>