Improving application timing predictability and caching performance on multi-core systems


Modern computing systems are constructed using commodity multi-core processors, on which part of the memory subsystem is shared by different cores on the same processor. Multiple applications executing simultaneously on a multi-core system contend for the shared memory resources, such as last level caches (LLC) and main memory, causing inter-application interference. Such inter-application interference, if uncontrolled, results in unpredictable execution delay for individual applications and severe system performance degradation. In this dissertation, we focus on shared cache interference and investigate two issues raised by the increasing complexity of underlying hardware and software for multi-core systems: timing predictability of real-time computing and caching performance for high performance computing.