The massive degree of parallelism inherent in SIMD computer systems has been exploited to produce high-performance and cost-effective solutions for many problems in the scientific and engineering communities. An SIMD system that can be used to accelerate important database queries will be developed. Initial results demonstrate that a performance improvement of 20 to 50 times over currently available technology and a cost/performance advantage of 100 times is attainable. The specific database queries that are accelerated are JOINs, Range SELECTs, and SCALAR Aggregates. These three classes of queries combine to underpin the decision support functions of large database installations (as opposed to the transaction processing functions). In addition to providing the potential for increased performance for existing database installations, the nested relational model, which the accelerator supports, can provide the basis for building database systems on complex structured objects such as image and object-oriented databases. The goal of this project is to integrate the accelerator technology developed into commercial relational database products.This project is applicable to satellite and telescope generated images, as well as commercially generated data such as supermarket scanner data that produce huge volumes of raw data.SIMD, relational database management system, nested relational modelSTATUS: Phase I Only
