Distributed dynamic hash tables using IBM LAPI
Proceedings of the 2002 ACM/IEEE conference on Supercomputing
An Efficient Caching Scheme for Software RAID File System in Workstation Clusters
HPC-ASIA '97 Proceedings of the High-Performance Computing on the Information Superhighway, HPC-Asia '97
| Hi-index | 0.01 |
This document describes an Active Message layer we have constructed on a Net work of Workstations. Active Messages is a thin, highly optimized communication layer targeted at the library or compiler writer. A primary goal of an Active Message layer is to deliver the minimum latency and peak bandwidth of the network hardware to user programs. Previous work on Active Messages demonstrated an order of magnitude performance improvement over vendor supplied send and receive libraries for Massively Parallel Processors. The HPAM layer currently runs on a network of 4 HP 9000/735 workstations with Medusa FDDI interface cards. HPAM achieves a round trip time for a 20 byte pay load of 29 msec, an order of magnitude better than traditional software, and a maximum bandwidth of 12 MB/s.