Performance Modeling for Helper Thread on Shared Cache CMPs

Abstract---In data intensive applications of Cloud Computing such as XML parsing, large graph traversing and so on, there are a lot of operations to access irregular data. These data need be timely prefetched into the shared cache in CMPs by helper thread. However, a bad prefetching strategy of helper thread will cause the multi-core shared cache pollution and degradation of performance. For analyzing the stand or fall of prefetching strategy of helper thread, this paper proposes a performance model for helper thread prefetching, then aiming at relevant to standard testing programs in Olden and CPU2006, the experimental results show that our model is effective for identifying the distribution of timely, late, bad, ugly of helper thread LLC requests. We could find the stand or fall of prefetching strategy of helper thread by observing the foremost factors in the distribution. Keywords--multi-core shared cache; Helper thread; Performance model I