In a processor that is running multiple programs, the last-level cache is typically shared by all the
Question:
In a processor that is running multiple programs, the last-level cache is typically shared by all the programs. This leads to interference, where one program’s behavior and cache footprint can impact the cache available to other programs. First, this is a problem from a quality-of-service (QoS) perspective, where the interference leads to a program receiving fewer resources and lower performance than promised, say by the operator of a cloud service. Second, this is a problem in terms of privacy. Based on the interference it sees, a program can infer the memory access patterns of other programs. This is referred to as a timing channel, a form of information leakage from one programto others that can be exploited to compromise data privacy or to reverse-engineer a competitor’s algorithm. What policies can you add to your last-level cache so that the behavior of one program is immune to the behavior of other programs sharing the cache?
Step by Step Answer:
Computer Architecture A Quantitative Approach
ISBN: 9780128119051
6th Edition
Authors: John L. Hennessy, David A. Patterson