User Tools

Site Tools


software:specfem

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
software:specfem [2018/03/05 15:00]
wphase
software:specfem [2018/03/05 16:06] (current)
wphase
Line 284: Line 284:
  
  
-==== Parallel SPECFEM3D_GLOBE RUNS ====+==== Running SEM simulations in parallel ​==== 
 + 
 +Some instructions to use custom scripts enabling parallel SEM simulations on the HPC cluster 
 + 
 +=== Preparing the input files ===
  
-To launch batch of SEM runs: 
  
 First, create an event list "​Events.txt"​ with 3 collumns: First, create an event list "​Events.txt"​ with 3 collumns:
-1st column: event_id (will also be the name of the run directory) +  * 1st column: event_id (will also be the name of the run directory) 
-2nd column: path to ''​CMTSOLUTION''​ file for this event +  ​* ​2nd column: path to ''​CMTSOLUTION''​ file for this event 
-3nd column: path to ''​STATION''​ file for this event (can be the same for all events)+  ​* ​3nd column: path to ''​STATION''​ file for this event (can be the same for all events)
  
-Then you must setup a ''​Par_file'' ​as described above (be careful to use a version of ''​Par_file''​ that is compatible with your SEM version)+Then you must setup a ''​Par_file''​ (be careful to use a version of ''​Par_file''​ that is compatible with your SEM version)
  
-The SPECFEM runs will be handled using 3 scripts: +Finally, you must setup hostfiles named ''​nodelistN'' ​files where N=0,​...,​Np-1 (Np, the number of parallel SEM simulations)These files must specify host names and number of slots per node. Here is an example: 
-''​parallelSEM.sh''​: is the main script +<​code>​ 
-- ''​run_gpu_nodelist.sh'' ​is the script used to run the mesher and solver +$ cat nodelist0  
-- ''​sleep.slurm'' ​is a script to reserve the GPU nodes+hpc-n443 slots=8 
 +hpc-n444 slots=8 
 +hpc-n445 slots=8 
 +</​code>​ 
 +(see ''​/​b/​home/​eost/​zac/​jobs/​specfem/​parallelSEM/​nodelist0''​)
  
-Before running ​it, make sure the input parameters in ''​parallelSEM.sh''​ are consistent with the input parameters stated above (see ''​INPUT PARAMS''​ in the main script). Then run:+=== Running the simulations in parallel === 
 + 
 +Parallel SEM simulations are handled using 3 scripts: 
 +  * ''​parallelSEM.sh'':​ is the main script, that compiles the code and run the simulations 
 +  * ''​run_gpu_nodelist.sh''​ is the script used to run the mesher and solver 
 +  * ''​sleep.slurm''​ is a script to reserve the GPU nodes 
 +All these scripts are available in ''/​b/​home/​eost/​zac/​jobs/​specfem/​parallelSEM''​ 
 + 
 +Before running ​your job, make sure that the input parameters in ''​parallelSEM.sh''​ are consistent with the input parameters stated above (see ''​INPUT PARAMS''​ in the main script). ​Specifically:​ 
 +  * ''​SPECFEMDIR'':​ path to SPECFEM3D_GLOBE directory 
 +  * ''​Par_file'':​ path to the Par_file used in simulations 
 +  * ''​Nparallel'':​ Number of SEM simulations in parallel (make sure enough GPUs are available) 
 +  * ''​event_list'':​ List of events with the format given above 
 +  
 +Then run your simulations:
 <​code>​ <​code>​
 ./​parallelSEM.sh ./​parallelSEM.sh
 </​code>​ </​code>​
 +The script will make sure that the GPU nodes are available before launching SPECFEM3D_GLOBE.
 +
 +
  
  
software/specfem.1520258418.txt.gz · Last modified: 2018/03/05 15:00 by wphase