This work describes the use of parallel programming technologies in molecular dynamics modeling programs. A review of hardware platforms is given, as well as the programs in which they are applied. On the example of the PUMA-CUDA modeling program, it is explained how technologies help accelerate the process of computational experiments. The main computational complexity is to calculate the non-valent interactions; graphic accelerators are used to calculate them. In the absence of such - multi-core processors according to the OpenMP technology. MPI technology is recommended for tasks that are not enough resources of one computing node.

molecular dynamics simulation programs, PUMA-CUDA, OpenMP, MPI, CUDA