Browse Prior Art Database

Snug Coupled Multiprocessor

IP.com Disclosure Number: IPCOM000085564D
Original Publication Date: 1976-Apr-01
Included in the Prior Art Database: 2005-Mar-02
Document File: 2 page(s) / 43K

Publishing Venue

IBM

Related People

Coleman, CD: AUTHOR

Abstract

A multiprocessor arrangement is illustrated in the figure and represents a cross between "loosely-coupled" multiprocessor systems requiring input/ output (I/O) to transfer data and direct coupled systems, which must be synchronized with extensive queues to avoid memory conflicts.

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 71% of the total text.

Page 1 of 2

Snug Coupled Multiprocessor

A multiprocessor arrangement is illustrated in the figure and represents a cross between "loosely-coupled" multiprocessor systems requiring input/ output (I/O) to transfer data and direct coupled systems, which must be synchronized with extensive queues to avoid memory conflicts.

Each processor 11-13 includes a main storage, a portion of which may be directly accessible by another processor through a "snug couple adapter" 14-16 connected as a channel. Although the adapter appears to the processor as a channel, it is actually only an interface adapter to interface between the processor channel socket and data transfer units 17-18.

An exemplary snug couple adapter would have nearly 250 interface lines and would be operable either in single-cycle mode to transfer from 1 to 8 bytes or in cycle-steal mode to transfer multiple double words.

The data transfer units each act as a channel, fetching and executing each channel command of a source processor, the processor executing an I/O instruction. Each processor is assigned a unique device address by which it can be addressed by any other processor. Multiple transmission paths are possible between processors to allow simultaneous transmission.

A source processor first opens input or output windows in its main storage for direct use by a target processor, the processor being addressed, and instructs the data transfer unit. A source processor can make direct references to main storage within the...