PERFORMANCE OF A HIGH-LEVEL PARALLEL LANGUAGE ON A HIGH-SPEED NETWORK

Clusters of workstations are often claimed to be a good platform for p arallel processing, especially if a fast network is used to interconne ct the workstations. Indeed, high performance can be obtained for low- level message passing primitives on modern networks like ATM and Myrin et. Most application programmers, however, want to use higher level co mmunication primitives. Unfortunately, implementing such primitives ef ficiently on a modern network is a difficult task, because their softw are overhead is relatively much higher than on a traditional, slow net work (such as Ethernet). In this paper we investigate the issues invol ved in implementing a high-level programming environment on a fast net work. We have implemented a portable runtime system for an object-base d language (Orca) on a collection of processors connected by a Myrinet network. Many performance optimizations were required in order to let application programmers benefit sufficiently from the faster network. In particular, we have optimized message handling, multicasting, buff er management, fragmentation, marshalling, and various other issues. T he paper analyzes the impact of these optimizations on the performance of the basic language primitives as well as parallel applications. (C ) 1997 Academic Press.