An object-oriented concurrent programming model for simulation applications on distributed memory processors network

We first explain our motivations to develop parallel architectures and software for realistic rendering and simulation applications. Then, after a brief summary of the results obtained with a previous prototype on a Transputer network with a classical programming approach (whose aim is to clearly explain the existing problems for the classical critical aspects of load balancing, deadlock, etc.), we explain why and how an object-oriented approach associated with an SPMD (single process, multiple data) programming model and its implementation appears to be well suited and more efficient. This demonstration is made on both sides of the problem: the data distribution and the communication model