Search

Users of parallel machines need to have a good grasp for how different communication patterns and styles affect the performance of message-passing applications. LogGP is a simple performance model that reflects the most important parameters required to estimate the communication performance of parallel computers. The message passing interface (MPI) standard provides new opportunities for developing high performance parallel and distributed applications. In this paper, we use LogGP as a conceptual framework for evaluating the performance of MPI communications on three platforms: Cray-Research T3D, Convex Exemplar 1600SP, and a network of workstations (NOW). We develop a simple set of communication benchmarks to extract the LogGP parameters. Our objective in this is to compare the performance of MPI communication on several platforms and to identify a performance model suitable for MPI performance characterization. In particular, two problems are addressed: how LogGP quantifies MPI performance and what extra features are required for modeling MPI, and how MPI performance compare on the three computing platforms: Cray Research T3D, Convex Exemplar 1600SP, and workstations clusters.

The processing and the communication capacities of interconnected workstations are currently not prohibitive for executing parallel and distributed applications. The programs can be executed on remote workstations yet deliver the results locally or where they are required, provided the overhead is tolerable. This capability opens up a whole new era. With a task (or process) migration facility, it would be possible to utilize abundant processing power and realize fail-proof execution of applications. This paper presents a task migration infrastructure (TMI), which allows tasks to dynamically migrate amongst the workstations. It incorporates parallel and distributed applications. Contrary to other studies, TMI is partly integrated into the UNIX kernel in the form of enhancements (E-kernel) and it is based on a dedicated message passing communication interface. TMI is to serve as an infrastructure for higher level usage such as fault tolerance, load balancing, and resource utilization. The experimental results have acknowledged the feasibility of TMI, in particular for long running applications.

Many large-scale multicomputers communicate through message passing. This makes the design of the underlying network is a crucial issue for multicomputers. Many interconnection networks have been proposed and studied, but none has been shown to be universally applicable. A sizeable performance improvement may be possible by combining the features of two or more types of interconnection network architectures. Also, as the system size increases, there is a locality of communication among the processors which can be exploited for performance gains. Hierarchical networks provide a means to achieve both these performance improvements. In this paper, we propose a new hierarchical fault-tolerant interconnection network that combines the hypercube and the torus. Simulation results show that the fault coverage and mean internodal distance for the proposed network are better than those achieved by both the hypercube and the torus.

LogGP is a simple parallel machine model that reflects the important parameters required to measure the real performance of parallel computers (Alexandrov et al., 1995). The message passing interface (MPI) standard provides new opportunities for developing high performance parallel and distributed applications. We use LogGP as a conceptual framework for evaluating the performance of MPI communications on three platforms: Cray-research T3D, Convex Exemplar 1600, and a network of workstations (NOW). Each machine is viewed as a "black box" that supports the same MPI implementation.