Implementation and Performance Evaluation of Locust

This paper presents the implementation and performance evaluation of Locust, a compiler-directed distributed shared memory system on a network of workstations. The thesis of the Locust project is that the distributed shared memory program performance can closely approximate message passing program performance if compiler-generated data dependency information is eeectively exploited. Locust uses this information to address the issues of false sharing, coherence maintenance, data prefetching, and aanity scheduling. Locust also takes a function-shipping approach to support operations on mutual exclusion objects and thus signiicantly reduces network traac. We have implemented the rst Locust prototype on a network of 90-MHz Pentium PCI machines running FreeBSD 2.0.5 connected by a 100 Mbps Fast Ethernet. For the four test parallel applications, Locust achieves a performance level that is within 5-10% of the message passing performance. In addition, Locust performs up to 23% faster than a comparable write-update protocol implementation.