Any views expressed within media held on this service are those of the contributors, should not be taken as approved or endorsed by the University, and do not necessarily reflect the views of the University in respect of any particular issue.
Abstract: To fully leverage modern hardware capabilities such as extensive parallelism and large caches, software must explicitly adapt to benefit from these features. Theoretically, operating systems could optimize application execution, for instance through intelligent scheduling. Practically, they lack essential insights about the application: “traditional” processing models like threads are a “black box” that hardly allow any information exchange between the application and the system underneath.
This talk presents MxTasking, a novel processing model that re-thinks the interfaces between applications and the execution substrate. The talk will delve into optimizations and simplifications made possible by this model. Additionally, we will focus in detail on CPU cache prefetching, one of MxTasking’s key optimizations, exploring its potential and associated challenges.
Comments are closed
Comments to this thread have been closed by the post author or by an administrator.