Difference between revisions of "Gkeyll Job Scripts"
Jump to navigation
Jump to search
| (2 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
#!/bin/bash | #!/bin/bash | ||
| − | #SBATCH --time | + | #SBATCH --time # wall time |
| − | #SBATCH --ntasks= | + | #SBATCH --ntasks= # number of mpi processes |
#SBATCH --cpus-per-task= # Number of OpenMP threads for each MPI process/rank | #SBATCH --cpus-per-task= # Number of OpenMP threads for each MPI process/rank | ||
#SBATCH --nodes= # Number of nodes | #SBATCH --nodes= # Number of nodes | ||
#SBATCH --ntasks-per-node= # How many tasks on each node | #SBATCH --ntasks-per-node= # How many tasks on each node | ||
#SBATCH --ntasks-per-socket= # How many tasks on each CPU or socket | #SBATCH --ntasks-per-socket= # How many tasks on each CPU or socket | ||
| − | #SBATCH --mem-per-cpu= | + | #SBATCH --mem-per-cpu= # Memory per core |
# load modules | # load modules | ||
| Line 15: | Line 15: | ||
# run the program | # run the program | ||
| − | srun --mpi=pmix_v3 gkyl | + | srun --mpi=pmix_v3 gkyl input.lua >& output.log |
| − | *Note: The user must decide the number of tasks and how they are distributed. This depends on | + | *Note: The user must decide the number of tasks and how they are distributed. This depends on on the availability of resources and the extent of the parallelization of the software. Also user must specify the the account ($SBATCH --account) and the Quality of Service (QOS) (#SBATCH --qos) must be specified. |
| − | on the availability of resources and the extent of the parallelization of the software. Also | ||
| − | user must specify the the account ($SBATCH --account) and the Quality of Service (QOS) (#SBATCH | ||
| − | --qos) must be specified. | ||
Latest revision as of 14:32, 3 June 2024
#!/bin/bash #SBATCH --time # wall time #SBATCH --ntasks= # number of mpi processes #SBATCH --cpus-per-task= # Number of OpenMP threads for each MPI process/rank #SBATCH --nodes= # Number of nodes #SBATCH --ntasks-per-node= # How many tasks on each node #SBATCH --ntasks-per-socket= # How many tasks on each CPU or socket #SBATCH --mem-per-cpu= # Memory per core
# load modules module load openmpi/4.1.5 module load openblas/0.3.24 module load superlu_dist/6.0.1 module load metis/5.2.
# run the program srun --mpi=pmix_v3 gkyl input.lua >& output.log
- Note: The user must decide the number of tasks and how they are distributed. This depends on on the availability of resources and the extent of the parallelization of the software. Also user must specify the the account ($SBATCH --account) and the Quality of Service (QOS) (#SBATCH --qos) must be specified.