A Survey of Three Scalable Hydrodynamics Codes

Massively parallel computer architectures offer greater memory capacity and faster computational speeds through the cooperative effort of hundreds or thousands of individual processors. This makes possible the solution of problems previously impractical with conventional vector supercomputers. This report presents a survey and partial assessment of several different hydrodynamic codes with a focus on scalability; i.e., codes that are amenable to implementation on high-performance parallel computer architectures. Issues, such as decomposing the physical problem into naturally parallel parts, load balancing the computation across multiple processors, effectively communicating data between processors, and manipulation and display of large sets of data are discussed from the perspective of three different hydrodynamic codes. Original contains color plates: All DTIC/NTIS reproductions will be in black and white.