2D KH test

[oporth]

• big enough to scale to a full node (128 cores on Snellius) - since we want to compare the performance of the full CPU node vs full GPU node later
• Use also as test case for hybrid parallelization

[oporth]

The test case is in tests/hd/AGILE_baseline its a 4096x4096 uniform grid hydro sim of a KH problem. But I run it only for 100 iterations.
Initial performance and scaling (flat MPI) has been carried out on Snellius with intel and nvhpc modules In particular

module load 2022
module load OpenMPI/4.1.4-NVHPC-22.7-CUDA-11.7.0

for nvhpc

respectively

module load 2022
module load impi/2021.6.0-intel-compilers-2022.1.0

for the intel modules.

The scaling plot is here: snellius_baseline.pdf raw performance data in the test folder.

I used the timer " Total timeloop took : XXX sec" for the performance measurement.