The Gromacs package is in a continuous process of enrichment with new algorithms for molecular dynamics. Going from the current level of prototyping for those algorithms to the level of professional parallelization requires a concentrated, guided effort. In this project we aim to implement and parallelize a number of promising recent algorithms for molecular dynamics in Gromacs. A second direction is to add to the existing mixed MPI-SMP scheme of parallelization the use of graphical processors through the involvement of GPU-CUDA programming. Many times the program is run on single quadcores or opteron stations and adding the possibility of using the power of graphical cards will speed up the simulations.