Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Method for Maximizing Productivity from a Computer Equipped with a Multiplicity of Central Processing Units

IP.com Disclosure Number: IPCOM000106054D
Original Publication Date: 1993-Sep-01
Included in the Prior Art Database: 2005-Mar-20
Document File: 2 page(s) / 84K

Publishing Venue

IBM

Related People

Johnson, WJ: AUTHOR [+4]

Abstract

Provided is a method for enhancing a programming language with directives for allowing program functions to be automatically directed to distinct Central Processing Units (CPUs).

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Method for Maximizing Productivity from a Computer Equipped with a Multiplicity of Central Processing Units

      Provided is a method for enhancing a programming language with
directives for allowing program functions to be automatically
directed to distinct Central Processing Units (CPUs).

      Programs developed for parallel processing systems are
developed by having the programmer organize source code and design so
that events which occur will maximize the efficiency of program
execution.  A method is needed which alleviates the programmer from
programming sophisticated interprocess communication for the purpose
of maximizing parallel processor computer hardware.

      For example, a scientific application may require three very
complicated and timely function executions where 1 execution is
dependent on the other 2 (i.e., one calls the other 2 functions).
The programmer has a choice of either programming a single simple
program where the two called functions, call them FUNCTION CALL1 and
FUNCTION CALL2, are called from the main program FUNCTION WIDGET.
However, if FUNCTION CALL1 takes one month of super-computer
execution time and FUNCTION CALL2 takes one month of super-computer
execution time, and FUNCTION WIDGET takes one month of super-computer
execution time then the entire program will take three months for
completion.  The programmer knows his computer is equipped with
multiple CPUs and an operating system which supports interprocess
communications.  Therefore, with extra programming skill and time,
the programmer is able to produce three programs which run on their
own CPUs concurrently, and then communicate at the correct
coordinated times in order to achieve an approximate 1 month
super-computer execution time.  The problem here is the programmer
spent substantial time coordinating and programming interprocess
communications between the three functions.  A method which will cut
down program development time for parallel processor environments is
desirable.

      This article defines a new programming language directive for
allowing program functions to be automatically directed to a distinct
CPU.  This allows functions to be directed to a CPU from a single
program (and source file) so that the functions are elaborated
co...