A Methodology for the Performance Modelling of Distributed Cache Coherent Multiprocessors
Anthony J. Field, Peter G. Harrison · 2024
In this chapter we present a general method for developing of performance models of shared-memory computer systems. These machines comprise a number of architecturally identical processing nodes whose activities are coordinated via a single global shared memory system. Each node has the ability to cache a part of the shared memory in separate high-speed stores local to each processor and the role of the coherency protocol is to ensure that each processor&s;s view of the shared memory is the same at every instant.