Running on HPC

Blackbox can also run on High Performance Computing (HPC) clusters and process to generate samples is same as running on local machine. If you encouter a shared memory error while running on HPC, it is recommended to use following command to launch your python file in slurm job script:

mpirun -n 1 --mca pml ob1 --mca btl self,tcp python runscript.py

Note

Only one processor is used to launch the python file. The number of processors used for meshing and flow solver is specified while initializing Blackbox object, refer sample generation section for more details.

This command changes the mode of communication between the processors to TCP which will avoid the shared memory error.