© 1963 by British Computer Society
Estimating computer performance
Auerbach Corporation, Philadelphia, Pa, USA
Methods are described and illustrated for measuring the performance of different computer systems and configurations on standardized problems, as they are being developed as part of Standard EDP Reports.
In order to provide reliable performance comparisons, the methods of measurement must be objective. Standardized problems must be precisely defined to minimize subjective error and be unbiased toward particular computer types, yet they should be flexible enough to exploit useful individual features.
One of the standard problems is file processing run in which records from a detail file are used to update a master file, and a printed record of each change is produced. Record layouts are fixed for the detail and report files but are left flexible for the master file. Detailed block diagrams define the internal operations that must be performed.
Graphs of performance corresponding to different parameters are presented and discussed. The results provide reliable performance measures, and comparable among many different computers and tasks, and are economical to produce.