KRS1

Abstract

Historically, shared memory and virtual memory have been the main line programming model from an applications and computer science perspective for two reasons: shared memory is a flexible and high performance means of communicating between processors, tasks or threads and; shared memory provides high programming efficiency through use of conventional memory management methods. The KSR1™ bridges the gap between the historical shared memory model and MPPs by delivering the shared memory programming model and all of its benefits, in a scalable, highly parallel architecture. The KSR1 runs a broad range of mainstream applications, ranging from numerically intensive computation, to on-line transaction processing (OLTP) and database management and inquiry. The use of shared memory enables a standards based open environment. The KSR1’s shared memory programming model is made possible by a new architectural technique called ALLCACHE™ memory. The shared memory programming model delivered by ALLCACHE facilitates porting and high performance for customer and third party applications. Within this context, the KSR1 architecture and applications environment consisting of a conventioanl UNIX based OS (symmetrically executing across all processors) and programming environment (including conventional languages) will be described. The performance and porting experiences of two applications on the KSR1 will be discussed as well as the Kendall Square feature called Query Decomposer which automatically parallelizes complex queries tunning under Oracle7.