Аннотация:The DVM system was designed to create parallel programs of scientific-technical calculations in CDVMH and Fortran-DVMH languages. These languages use the same model of parallel programming (the DVMH model) and are extensions of standard C and Fortran languages by parallelism specifications, implemented as compiler directives. The DVMH model allows to create efficient parallel programs for heterogeneous computational clusters, which nodes use as computing devices not only general purpose multi-core processors but also can use attached accelerators (GPUs or Intel Xeon Phi coprocessors). This article
discusses new possibilities to work with irregular grids and graphs, which were implemented in the CDVMH compiler recently. Using the developed extension can considerably simplify a parallelization of irregular grid applications on a cluster.