sci-chemistry ottxor@gentoo.org Christoph Junghans Enable cuda non-bonded kernels More precise calculations at the expense of speed Single precision version of gromacs (default) Enable building of Fortran Kernels for platforms that dont have assembly loops Enable sse4.1 acceleration Enable 128bit avx with fma (e.g. AMD BullDozer) Enable 256bit avx (e.g. Intel Sandy Bridge) Add acceleration through sci-libs/openmm Enable Power6 optimizations