Implementation of atomic primitives on distributed shared memory multiprocessors

Many hardware primitives have been proposed for synchronization and atomic mem-ory update on shared-memory multiprocessors. In this paper, we focus on general-purpose primitives that have proven popular on small-scale bus-based machines, but have yet to become widely available on large-scale, distributed-memorymachines. Specif-ically, we propose several alternative implementations of fetch and , compare and-swap, and load linked/store conditional. We then analyze the performance of these implementations for various data sharing patterns, in both real and synthetic applica-tions. Our results indicate that good overall performance can be obtained by imple-menting compare and swap in a multiprocessor's cache controllers, and by providing an additional instruction to load an exclusive copy of a line