Parallelization of SPHC on the Intel iPSC/860
T.P. Hughes, A.J. Giancola · Bulletin of the American Physical Society · 1993
SPHC is a C-language implementation of smoothed-particle hydrodynamics. The authors are currently modifying the code to run on the 16-node Intel iPSC/860 at Phillips Laboratory. Orthogonal recursive bisection (ORB) is used to spatially decompose the simulation volume onto the processors. The short particle interaction range significantly reduces the volume of interprocessor communication compared to astrophysical N-body particle codes. Use of variable smoothing-length simplifies the load-balancing problem since it tends to keep the number of neighbors per particle constant. The authors use the Portable Instrumented Communication Library for inter-processor message-passing. The parallel performance of the code is described.