Fast Primitives for Irregular Computations on the NEC SX-4
M. Ballabio, M. Boverat, M.F. Hodous, Drago Marić · 1997
Modern science and engineering disciplines make extensive use of computer simulations. As these simulations increase in size and detail, the computational costs of naive algorithms can easily become prohibitive. Fortunately, sophisticated modeling techniques have been developed in many areas that that vary model resolution as needed coupled with sparse and adaptive algorithms that vary computational effort in time and space as needed. For example, adaptive spatial decompositions are used with fast n-body techniques such as FMA or PM in cosmological simulations, and adaptive unstructured meshing and sparse linear system solvers are used in computational fluid dynamics. The computations that result from these techniques are called irregular, reflecting the a priori unpredictability in the reference pattern and distribution of work.