Analysis of parallel processing effects through a co-processor in diesel engine management system

Abstract

We show that computational load can be reduced when a programmable co-processor is used for a time-critical system design. We analyze the real-time performances of a diesel engine management system by applying a programmable timer co-processor. In order to verify the performances caused by the co-processor, we investigated the timing parameters such as the process time, execution period, and processor utilization, according to a changing task structure. The processor utilizations of the tasks were found to be reduced within the range of 0.65% to 33.68% on worst-case execution period condition. These results show that an appropriate task construction considering the co-processor characteristics can lead to improvement in real-time performance from parallel processing on par with a multicore processor system even under limited conditions.