A new parallel TreeSPH code

Abstract

In this report we describe a parallel implementation of a Tree-SPH code realized using the SHMEM libraries in the Cray T3E supercomputer at CINECA. We show the result of a 3D test to check the code performances against its scalar version. Finally we compare the load balancing and scalability of the code with PTreeSPH (Davé et al 1997), the only other parallel Tree-SPH code present in the literature.