Event-Driven Monitoring of Parallel Systems

The dynamic behavior of parallel programs can be disclosed by event-driven monitoring. A hybrid monitoring method for analyzing the functional behavior and the performance of programs in parallel and distributed systems is presented. The paper contains a description of a scalable hardware monitor system (ZM4) and a software package for event trace evaluation (SIMPLE) which make our concepts available to programmers, assisting them in debugging and tuning of their code. A short survey of related monitor systems highlights the distinguishing features of our implementation. As an application of our monitoring and evaluation system, the analysis of a parallel ray tracing program running on the SUPRENUM multiprocessor is described. Models can help to increase the efficiency when writing parallel programs. We show that monitoring and modeling both rely on a common abstraction of a system's dynamic behavior and therefore can be integrated to one comprehensive methodology. This methodology is ..