Scalable parallel fault simulation for shared-memory multiprocessor systems
Stavros Hadjitheophanous, Stelios Neophytou, Maria K. Michael · 2016
Multicore architectures can significantly accelerate the performance of well-established design and test automation processes, provided that the underlying process is scalable with respect to the system on which it is executed. In this work we concentrate on fault simulation and propose a new parallel process for shared-memory multicore systems, capable of maintaining its scalability as the number of processing cores utilized increases. In order to maximize parallelization, the method utilizes a simple, non-optimized single thread simulation process, which allows for high degrees of freedom to be exploited by three different and combined dimensions of parallelism. Simulation data is distributed to the available cores in a balanced fashion in order to favor speed-up over single-core executions and, ultimately, scalability. The experimental results show that the proposed approach achieves high speed-up rates which, in contrast to comparable state-of the-art methods, increase monotonically with the number of cores demonstrating a highly scalable solution.