Abstract
In modern computer hardware, increased processing capacity is gained by introducing more processor cores, not with increased clock speed. For individual cores this may imply reduced processing capabilities. Consequently, applications that are not designed to benefit from multiple cores need to be partially re-engineered for maintaining the existing performance. Such revisit can be eased with libraries that support concurrency. In this paper, we introduce our experiences on revisiting the design of an industrial system with multicore processing in mind, as well as on verifying the effect of our improvements with performance tests.