Skip to main content
. 2020 Sep 29;21:424. doi: 10.1186/s12859-020-03711-2

Fig. 1.

Fig. 1

Hybrid OpenMP/MPI implementation of FVA ensures two levels of parallelism. The distribution of tasks is implemented following a hierarchical model where MPI manages coarse-grained parallelism in non-shared memory systems. At a lower level, OpenMP processes within each MPI process manage fine-grained parallelism taking advantage of the shared memory to improve performance