Changes between Version 1 and Version 2 of DocBenvAidrisAada


Ignore:
Timestamp:
03/24/14 16:14:53 (6 years ago)
Author:
trac
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • DocBenvAidrisAada

    v1 v2  
     1{{{ 
     2#!html 
     3<h1>Working on Ada</h1> 
     4}}} 
     5---- 
     6[[PageOutline(1-3,Index du chapitre,,numbered)]] 
     7 
     8## IDRIS users' manual ## 
     9 * See: http://www.idris.fr/ada/ for ada 
     10 * See: http://www.idris.fr/adapp for adapp : pre-post-treatment 
     11  
     12## Commands to manage jobs on ada  ##  
     13 * The job's time limit is measured in real time, for example 1 hour on 32 procs accounts for 32 hours. Be careful not to have too much time on 1 processor. 
     14 * llsubmit --> submit a job 
     15 * llcancel --> cancel a job  
     16 * llq -u ''login'' --> indicates all jobs in the queue or running for the login ''login'' 
     17 * Trick: parameterize the llq display to see the job names 
     18{{{ 
     19llq -u $(whoami) -f %jn %id %st %c %dq %h -W 
     20}}} 
     21 * Post-mortem : jar , jar -l -j #jobid#, to obtain detailed information: memory, real time, efficiency,... 
     22  * on adapp, use : `/smplocal/bin/jar` 
     23 * Example of jar output : 
     24{{{ 
     25ada > jar 
     26|----------------------------------------------| 
     27|--- IDRIS/CNRS. Version du 18 février 2013 ---| 
     28|----------------------------------------------| 
     29 
     30Sorties concernant l'identifiant rpslxxx pour la période du 
     31        ==> 01 juin 2013 au 19 juin 2013 
     32 
     33 
     34 Owner   Job Name        JobId      Queue tEse  tCpu   #T   (%)   S 
     35------- ----------- --------------- ----- ---- ------ --- ------- - 
     36rpslxxx ADA337      ada338.290170.0 c32t2  133   1232  32   28.95 C 
     37rpslxxx ADA337      ada338.290333.0 c32t2 5425 165141  32   95.13 C 
     38rpslxxx PACKDEBUG   ada338.290610.0 t2      11      2   1   18.18 C 
     39rpslxxx ADA337      ada338.290438.0 c32t2 5471 166878  32   95.32 C 
     40rpslxxx PACKRESTART ada338.290611.0 t2     182     25   1   13.74 C 
     41rpslxxx REBUILDWRK  ada338.290612.0 t2    1577    503   1   31.90 C 
     42rpslxxx PACKOUTPUT  ada338.290730.0 t2     114     43   1   37.72 C 
     43}}} 
     44 
     45## Example of a job to start an executable in MPI ## 
     46Here is an example of a simple job to start an executable orchidee_ol (or gcm.e commented).  
     47The input files and the executable must be in the directory before starting the executable. 
     48{{{ 
     49#!/bin/ksh 
     50# ###################### 
     51# ##   ADA IDRIS   ## 
     52# ###################### 
     53# Query's name 
     54# @ job_name = test 
     55# Job type 
     56# @ job_type = parallel 
     57# Standard output file 
     58# @ output = Script_Output_test 
     59# Error output file (the same) 
     60# @ error = Script_Output_test 
     61# Number of requested processes 
     62# @ total_tasks = 8 
     63# max. CPU time per MPI process hh:mm:ss 
     64# @ wall_clock_limit = 1:00:00 
     65# Number of task OpenMP/pthreads per MPI process 
     66### @ parallel_threads = 4 
     67# End of header 
     68# @ queue 
     69 
     70export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/smplocal/pub/NetCDF/4.1.3/lib:/smplocal/pub/HDF5/1.8.9/seq/lib 
     71 
     72poe ./orchidee_ol 
     73#poe ./gcm.e 
     74}}} 
     75 
     76## Information on Gaya files from Ada ## 
     77 
     78The `mfls` command on Ada provides information on the Gaya files. 
     79 
     80## Specificities libIGCM on Ada ## 
     81 
     82 
     83At IDRIS and for Ada, output files are 'packed' using libIGCM_v2, i.e. they are grouped by periods (in general 1 year) using the command `tar` or `ncrcat` for NetCDF output files. 
     84[[BR]] 
     85 
     86This has been a default setup at TGCC for a few months. It is a new feature since February 2013 for IDRIS. 
     87[[BR]] 
     88 
     89The diagram below shows the different options offered by libIGCM. The 3rd option is currently activated by default at IDRIS. This option implies that files must be temporarily stored on the $WORKDIR space, which means that a large storage is needed (at least 20 To).[[BR]] 
     90 
     91[[Image(libIGCM_options.jpg, 50%)]] 
     92 
     93The diagram below details the added jobs `pack_debug`, `pack_restart` and `pack_output` as well as the directories those jobs are using. Note that the files are temporarily stored in the $WORKDIR/IGCM_OUT directories before being grouped and sent on Gaya in the IGCM_OUT directories.[[BR]] 
     94 
     95[[Image(libIGCM_pack.jpg, 50%)]] 
     96 
     97You will obtain annual output files with 12 monthly values in the Output/MO directory if you put `PackFrequency=1Y` in `config.card`. This is the default grouping period of most configurations but you can of course change it. 
     98[[BR]] 
     99 
     100What you must remember: 
     101 * The tool [wiki:DocGmonitor#RunChecker RunChecker.job] is meant to help you monitoring your simulations. It offers a synthetic view of the different post processing jobs' status. 
     102 * The tool [wiki:DocGmonitor#Méthodeconseilléeencasdeproblèmedenvironnementcalcul clean_year.job] is meant to help you clean until the last successfully computed pack period. 
     103 * If you detect anomalies and must rerun part of the simulation, you will have to make new complete pack periods (e.g. filling a gap by running 1 month of simulation is out of the question).  
     104 * The restart files are stored and grouped on Gaya in the directory IGCM_OUT/.../RESTART 
     105 * The different output text-files are stored and grouped on Gaya in the directory IGCM_OUT/.../DEBUG 
     106 * The listings for pack-jobs outputs stay on Ada in the directory $WORKDIR/IGCM_OUT/.../Out 
     107 * If you put the `SpaceName=TEST`parameter in `config.card` the pack jobs will not be started and your simulation will be stored as before in the WORKDIR/IGCM_OUT directory. This can be very useful for short tests. 
     108 
     109To learn more about this Section, you can read the documentation on [wiki:DocFsimu Simulation and post-processing] and on [wiki:DocGmonitor Monitor, debug and relaunching.][[BR]] 
     110 
     111Finaly, in case of panic, visit us or send your questions to the list Home-users. 
     112