A note on parallel efficiency of fire simulation on cluster

On efficiency of fire simulation realization: parallelization with greater number of computational meshes

“…Similarly to [2,3] we use a rectangular 100 kW fire source in a corridor with dimensions of 7. The mesh sensitivity study performed in [2,3] indicates that the chosen mesh can be considered to be fine enough already since the 2 nd s to the fire.…”

“…The mesh sensitivity study performed in [2,3] indicates that the chosen mesh can be considered to be fine enough already since the 2 nd s to the fire. The total execution time of the sequential simulation (denoted by 1M) performed on a single computational core of the HPC cluster was about 17.45 hours.…”

“…The exact choice on how MPI processes will be mapped by particular computational nodes and subsequently bound to particular sockets or cores is provided by the mpirun script which is a part of Open MPI. In [2,3], we studied the sequential simulation described in the previous section as well as 9 parallel simulations parallelized by the parallel MPI model using a single computational node of the HPC cluster. We illustrated efficiency of two strategies for allocation of computational resources: --mapby core, --bind-to core and --map-by socket, --bind-to socket for parallel simulations with m≤12 [3].…”

“…In order to avoid interference of the tested simulation with other tasks being solved on the cluster, we reserved the whole computational nodes for the tested simulation exclusively. Similarly to [2,3], we decomposed the computational domain in the x-direction into m = 18, 24, 30, 36 cube computational meshes (denoted by 18M, 24M, 30M, 36M ) with the 2 cm mesh resolution (Table 1 and Figure 3). As the chosen mesh resolution and the simulated fire scenario were the same as in [2,3], the mesh sensitivity study performed in [3] for both the sequential and parallel simulations indicates that the chosen mesh resolution can be considered as fine enough.…”

“…It supports four programming models: sequential model designed for sequential computers, parallel MPI (Message Passing Interface) model designed for distributed memory systems, multithreading OpenMP (Open Multi-Processing) model designed for shared memory systems and hybrid MPI&OpenMP model designed for distributed shared memory systems. In our previous research we studied the impact of parallelization on efficiency and accuracy of a corridor fire simulation [2,3] using a single computational node of the HPC cluster at the Institute of Informatics of Slovak Academy of Sciences in Bratislava (Slovakia). The research focused on the parallel MPI model of FDS and on testing the main strategies of allocation of available computational resources to given parallel calculations.…”