Browse Prior Art Database

Solution for Shared/Private Memory for Multisystem Configuration

IP.com Disclosure Number: IPCOM000049506D
Original Publication Date: 1982-Jun-01
Included in the Prior Art Database: 2005-Feb-09
Document File: 2 page(s) / 63K

Publishing Venue

IBM

Related People

Tan, KG: AUTHOR

Abstract

A two multiple virtual system (MVS) under a single system image (I/O, are shared by all processors) is set forth. Under such a system, it is necessary to partition the main storage into private and shared regions. The hardware mechanism which controls the private/shared storage and performs the necessary broadcasting and cross-interrogate functions increases the performance of a four-way multiprocessor (MP) system and can have from 2.4 x unit processor (UP) to approximately 3.4 x UP performance.

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 53% of the total text.

Page 1 of 2

Solution for Shared/Private Memory for Multisystem Configuration

A two multiple virtual system (MVS) under a single system image (I/O, are shared by all processors) is set forth. Under such a system, it is necessary to partition the main storage into private and shared regions. The hardware mechanism which controls the private/shared storage and performs the necessary broadcasting and cross-interrogate functions increases the performance of a four-way multiprocessor (MP) system and can have from 2.4 x unit processor (UP) to approximately 3.4 x UP performance.

A four-way MP performance can be improved by operating two MVS systems under a single system image. In other words, each attached processor (AP) (one system controller (SC) and its connected processors) is under the control of one MVS. In this configuration, shared system storage must be provided for the shared system information between the two MVS systems. The overall system performance is expected to approach 2 x 1.7=3.4 x UP performance. This is primarily based on the fact that data cross-interrogations will, to a large extent, reduce to that between two processors and, because of the private storage for such AP, the overall storage contentions should be reduced, as well.

The key criterion to implement the above solution relies on the ability to partition the system storage into private and shared regions. The "private" region is private to a group of processors (in this case, private to processors connected to the same SC). Data cross interrogations will only be performed between the...