Storage of Numeric Data in ADAMS

Author:Haddleton, Russell, Institute for Parallel ComputationUniversity of Virginia

This paper presents a scheme for the representation and storage of numeric data. The implemention of this scheme within the ADAMS Database System solves the major problem of maintaining numeric data in a bitwise ordered form, while providing the additional benefits of simplicity, expandability, portability, and compactness. 1. The Problem There are many commonly employed formats for representing numeric data. The C language provides seperate types for the representation of integer and real data, and extends these float and int types to double and long in order to allow the language user access to a larger range of values and to a greater degree of precision. Another format involves simply storing numeric digits as characters (using, for example, ASCII or EBCDIC ) . This allows for a flexible degree of precision and magnitude, but is not efficient in terms of storage. Some systems rely on one format for in-memory manipulation, which we will call host format, and another for disk or tape storage, which we will call storage format. Criteria for successful in-memory representation and for persistent storage formats of numeric data depend on what is required of the system employing them. There are a number of goals and trade-offs. A. Goals
Note: Abstract extracted from PDF text

All rights reserved (no additional license for public reuse)
Source Citation:

Haddleton, Russell. "Storage of Numeric Data in ADAMS." University of Virginia Institute for Parallel Computation Tech Report (1992).

University of Virginia, Institute for Parallel Computation
Published Date: