Shasta: A System for Supporting Fine-Grain Shared Memory Across Clusters.
Daniel J. Scales, Kourosh Gharachorloo · 1997
Shasta is a system that supports shared memory in software on clusters of computers with physically distributed memory. A unique aspect of Shasta compared to most other software distributed shared memory systems is that shared data can be kept coherent at a fine granularity. In addition, the system allows the coherence granularity to vary across different shared data structures in a single application. Shasta implements the shared memory by transparently adding extra checking code at loads and stores in an application executable. The added code checks to see if the data is available locally and communicates with other processors if necessary. In this paper, we describe the design and programming interface of Shasta and report parallel performance results on our cluster, including the effect of optimizations such as varying the coherence granularity of important data structures. 1 Introduction As microprocessor performance has continued to increase rapidly in the past decade, workstati...