Abstract
Statistical routines written for general use often require large data storage arrays. A program using three n x n arrays, for instance, would require 30,000 words of memory for 100 variables. Usually the number of variables in any analysis is less than the maximum number of variables dimensioned in the program. This generally results in a number of unused data storage locations and a waste of computer resources. It would be advantageous to allocate storage for data arrays at execution time rather than at compile time, thereby minimizing storage requirements and turn around time for each program run.
A series of multivariate analysis programs described by Cooley and Lohnes (1971) have been rewritten incorporating a subroutine, CORE, which dynamically allocates memory in blank common during execution according to parameters specified on the first cards preceding the data deck. In multiple problem runs, CORE is capable of either reducing or increasing field length, depending upon the number of variables included in the analysis. As an example, the CORREL program in the Cooley and Lohnes package requires a field length of 43000, words to execute on the CDC 6400 under SCOPE 3.3 without the CORE routine. As few as 16000, words are needed with the CORE routine.
Article PDF
Avoid common mistakes on your manuscript.
Reference
Cooley, W. W., & Lohnes, P. R.Multivariate data anlysis New York: Wiley, 1971.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Derby, P.L. Statistical programs utilizing dynamic allocation of memory on the CDC 6400. Behav. Res. Meth. & Instru. 6, 59 (1974). https://doi.org/10.3758/BF03200298
Received:
Accepted:
Issue Date:
DOI: https://doi.org/10.3758/BF03200298