Skip to content

Latest commit

 

History

History
60 lines (35 loc) · 1.5 KB

README.md

File metadata and controls

60 lines (35 loc) · 1.5 KB
Raw

HPC HYDRO

This is a work repository for HPC1b.

Contributors: Mihai Tomozeiu, Christian Reinhardt, Rafael Kung.

how to compile & run

on your local university desktop opensuse

cd hpc/HYDRO_C/Src
make clean
make lmpi
cd ../Output
mpirun -n 16 ../Bin/hydro_lmpi -i ../Input/debug.nml

The binary is in HYDRO_C/Bin. You can use mpirun to run the program, or use HYDRO_C/Bin/run, which runs the command above.

on a cluster

We store the outputs from the simulations in ../Output/.

How to run a simulation:

There are in principe two possibilities:

  1. Allocate one (or more) nodes with salloc
  2. Submit a slurm batch job with sbatch

ZBOX:

  1. Allocate with salloc --partition=zbox and run with srun -n cores hydro_lmpi
  2. sbatch example-zbox.job (see example-zbox.job in this directory for details)

Important: Always compile with make lmpi and run hydro_lmpi!

Dora:

  1. Allocate with salloc (no partition) and run with aprun -n cores hydro_mpi
  2. sbatch example-dora.job (see example-dora.job in this directory for details and be sure to change ACCOUNT to uzh8)

Important: Always compile with make mpi and run hydro_mpi!

Generating plots and movies with multi_output_to_movies.py

###proper call:

python multi_output_to_movies.py (path to output directory) (number of steps) (delay for movie frames)

###example:

python multi_output_to_movies.py ~/git/HPC_Hydro/HYDRO_C/Output/Input_sedov_100x10 10 100

.png and .mpg files will be created in the same directory as the analysed output files