Following SPECFEM2D manual "Chapter 2"

1. Download SPECFEM2d with git clone https://github.com/geodynamics/specfem2d.git.
2. Go to specfem2d directory and type ./configure FC=gfortran CC=gcc MPIFC=mpif90 --with-mpi. If using ifort, specify with ./configure FC=ifort MPIFC=mpif90 --with-mpi.
   NOTE: If an error occurs, check FLAGS_CHECK in Makefile after ./configure ....
   It worked without any error with macOS Catalina 10.15.7; GNU Fortran (GCC) 8.2.0; Apple clang version 12.0.0 (clang-1200.0.32.2); MPICH Version: 3.2.1
3. Type make to compile executable files.

• 200km * 200km 2D full space P-SV elastic wave propagation.
• Total NSTEP = 5000, timestep = 7.0e-3 [s].
• Number of control nodes per element = 4
• Grid size = 250m
• Number of element = 640000
• Total number of grid point = 10250000
• PML absorbed boundary condition.
• Point source at the center of medium

1. Copy the example project Benchmark_fullspace into specfem2d/EXAMPLES/. You can remove the other examples, but please put it into EXAMPLE as the script uses relative paths.

2. Go to specfem2d/EXAMPLES/Benchmark_fullspace/ and set number of processors in DATA/Par_file, line 20: NPROC=1.

3. Run sh run_benchmark.sh.
   This shell script conducts the meshing and then computes wave propagation.

4. To plot waveform to validate your calculation, go to specfem2d/EXAMPLES/Benchmark_fullspace/plot_result and run python validate_waveform.py 1, where the number indicates the number of processors used like OUTPUT_FILE_NP1.
   NOTE: You need the following python modules: numpy (1.19.1); pandas (1.1.3);scipy (1.5.2). Versions are not strict, but please check it if your module is too old.

5. specfem2d/EXAMPLES/Benchmark_fullspace/plot_result/plot_cputime.py is a script to plot scalability of parallelization. Please note the number of cores and associated cputime from output (stdout), and add it into the script. You can plot with python plot_cputime.py