Browse Prior Art Database

Multiprocessing Scheduler

IP.com Disclosure Number: IPCOM000077987D
Original Publication Date: 1972-Oct-01
Included in the Prior Art Database: 2005-Feb-25
Document File: 2 page(s) / 61K

Publishing Venue

IBM

Related People

Chroust, G: AUTHOR

Abstract

In a multiprocessing environment the sequencing constraints between individual processes of a task are often specified by partial ordering (FORK/JOIN type structure as in Fig. 1), which is fixed throughout the computation of the task and indicates which processes have to be terminated before other processes can be initiated. It is known to represent this sequencing information in a matrix where, e.g., a "1" in the ith row and jth column indicates that the initiation of the ith process has to wait until the jth process is terminated. After a process has been terminated, the corresponding column is set to zeros. Subsequently, any process associated with a row of all zeros can be initiated.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 61% of the total text.

Page 1 of 2

Multiprocessing Scheduler

In a multiprocessing environment the sequencing constraints between individual processes of a task are often specified by partial ordering (FORK/JOIN type structure as in Fig. 1), which is fixed throughout the computation of the task and indicates which processes have to be terminated before other processes can be initiated. It is known to represent this sequencing information in a matrix where, e.g., a "1" in the ith row and jth column indicates that the initiation of the ith process has to wait until the jth process is terminated. After a process has been terminated, the corresponding column is set to zeros. Subsequently, any process associated with a row of all zeros can be initiated.

This method of multiprocessing sequencing control is improved by using a four-state functional memory which avoids changing the contents of the sequencing matrix and allows a fast recognition of processes ready for initiation. The sequencing constraints are stored in the four-state functional memory (associative memory) which accounts for the efficiency of the method. Information about the current state of the processes is retained only in I/O register of the functional memory, so that the functional memory itself can be shared by several independent tasks and could be established as a read-only storage. Each independent task would merely have its individual I/O register.

The only addition to a conventional functional memory is that the Y-state, forcing...