This paper examines the validity of Grosch’s law [1] and then goes on to examine three important studies related to evaluating the performance of computers. These include one undertaken by Knight in the early 1960s [2,3]; a second major study by Solomon in the mid-1960s [4]; and, finally, one by Cale et al. completed in the late 1970s [5].

The author then presents his methodology for the measurement of computer power using six variables, including maximum memory, maximum channels, cycle time, relative performance, MIPS, and price. The 106 systems used in this analysis were grouped into five categories which included super computers, large mainframes, small mainframes, minicomputers, and microcomputers. The author then shows that Grosch’s law holds within each category. He postulates that an organization should select the most powerful computer within each category. This is a well-written paper in an area which has received very little attention during the last 25 years.