Dependable execution of distributed programs

摘要

We report on the design and simulation of novel algorithms which ensure that application software runs correctly on a MIMD system in which processing units (PUs) can fail. The effect of these algorithms is evaluated for random task graphs using simulation. The simulation results are also compared to approximate analytical results. A specific application is finally studied: the Fast Fourier Transform. We give the corresponding task graph and then simulate its execution under various failure rates.