Browse Prior Art Database

STARLET AND THE APL MACHINES

IP.com Disclosure Number: IPCOM000128523D
Original Publication Date: 1975-Dec-31
Included in the Prior Art Database: 2005-Sep-16
Document File: 9 page(s) / 37K

Publishing Venue

Software Patent Institute

Related People

H. Berg: AUTHOR [+4]

Abstract

A brief description of the STARLET computer concept as well as of the APL machines of Abrams and Marchal is given. STARLET (which is a hardware concept) is compared with the API. machines (which are software concepts for APL interpreters). It is shown that all three systems differ essentially in all pertinent details such as the internal information structure and its interpretation, the function and scope of descriptors, the access and processing of data, etc. Furthermore, it is shown that the pseudo variable mechanism and the related condensation technique in STARLET is a unique method by which similar goals can be accomplished as by Abrams' beating and Marshal's deferred value computation method. It will be seen that the combination of Abrams' drag-along principle and the STARLET hardware organization would lead to a most efficient APL machine.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 10% of the total text.

Page 1 of 9

THIS DOCUMENT IS AN APPROXIMATE REPRESENTATION OF THE ORIGINAL.

STARLET AND THE APL MACHINES

by

W, K. Giloi and H. Berg

Department of Computer Science

Institute of Technology

University of Minnesota

Minneapolis, Minnesota 55455

University of Minnesota

Technical Report 75-20 July 1975 Cover courtesy of Ruth and Jay Leavitt

STARLET AND THE APL MACHINES by

W.K. Giloi and H. Berg University of Minnesota

Abstract

A brief description of the STARLET computer concept as well as of the APL machines of Abrams and Marchal is given. STARLET (which is a hardware concept) is compared with the API. machines (which are software concepts for APL interpreters). It is shown that all three systems differ essentially in all pertinent details such as the internal information structure and its interpretation, the function and scope of descriptors, the access and processing of data, etc. Furthermore, it is shown that the pseudo variable mechanism and the related condensation technique in STARLET is a unique method by which similar goals can be accomplished as by Abrams' beating and Marshal's deferred value computation method. It will be seen that the combination of Abrams' drag-along principle and the STARLET hardware organization would lead to a most efficient APL machine.

Keywords and phrases: Computer architecture, ordered sets as primitive data items, string and array processing, APL-type machine language, APL machines, APL interpreter, tagged architecture and descriptors, storage access functions, internal information structures

INTRODUCTION

STARLET is a novel concept of a hardware architecture, based on the introduc-tion of ordered sets as the primitive data structure in the machine, [1], [2], [3]. Consequently, the hardware processing unit deals.with ordered sets as entities (as contrasted to single components of such sets). In order to maximize the product of performance-to-cost ratio and flexibility, such entities are processed in a streaming mode of operation by a pipeline. To this end, the static data

University of Minnesota Page 1 Dec 31, 1975

Page 2 of 9

STARLET AND THE APL MACHINES

structures as stored in the memory have to be converted into respec-tive data streams. This conversion is accomplished by executing repetitively a general memory access function, resulting in memory address streams which control the flow of the data streams to and from the memory. By using homogeneous, rectang-ular arrays as the static representation of ordered sets, the memory access function is kept as simple as possible. This allows on-line execution during' the processing phase by a relatively inexpensive hardware and, hence, no access sets (sets of memory addresses) need be precalculated and stored. There is a natural classification of operations performed on ordered sets into two main categories, namely (1) operations which create only new structures but no new data and (2) operations which create new data as well as new structures. In the first category we...