Browse Prior Art Database

Highly Modular General Purpose Data Processing System

IP.com Disclosure Number: IPCOM000081531D
Original Publication Date: 1974-Jun-01
Included in the Prior Art Database: 2005-Feb-28
Document File: 4 page(s) / 50K

Publishing Venue

IBM

Related People

Jenny, CJ: AUTHOR [+2]

Abstract

A number of processor modules and other modular functional units, hereinafter called engines, are interconnected by a network of loops. Information is transferred and handled in blocks (frames) which are called External State Vectors (ESV). Clocks C1, C2, C3 and C4 release time frames for ESV's at regular intervals to their associated loops.

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 43% of the total text.

Page 1 of 4

Highly Modular General Purpose Data Processing System

A number of processor modules and other modular functional units, hereinafter called engines, are interconnected by a network of loops. Information is transferred and handled in blocks (frames) which are called External State Vectors (ESV). Clocks C1, C2, C3 and C4 release time frames for ESV's at regular intervals to their associated loops.

The loops provided are: Main Loop ML, Feeder Loop FL, Bypass Loop BL, Program Store Loop PSL, External Signal Loop ESL, and Instruction Fetch Loop IFL. The engines provided are: -ESV Create and Destroy Engine ESV-C/D, Instruction Fetch Engines IF, Instruction Processing Engines IP, Quiesced ESV Management Engines Q-ESV, Program Store Engine PS, Data Memory Management Engines DM-MGT, and I/O Attachment Engine I/O-ATT. Engines are small, simple processors. The number of engines of each class and of each type available on the system depends on the processing requirements.

Instruction processing engines are of several types, each type being capable of executing one instruction type (or a small set of instruction types).

Instruction Fetch Engines are responsible for access to the correct part of user programs stored in program Memory. Programs are initially stored by the Program Store Engine in cooperation with the I/O Attachment Engine. There exists a direct path between Program Store Engines and Instruction Fetch Engines, the Program Store Loop.

Data Memory Management Engines are directly coupled to memory modules called Data Memory or Global Memory (explained below). They are responsible for managing the attached memory (segmenting, partition-chaining, garbage collection, etc.) and for interpreting the memory address of the requesting ESV.

External input/output devices ED-1....ED-N are shown attached via an External Device Attachment Loop EDAL to the I/O Attachment Engine. This loop is not common to the loop network. It could be replaced with other transfer means interconnecting the external devices and I/O Attachment Engine.

Memory Modules: there are five classes of memory modules incorporated in the system: Program Memory PM, Data Memory DM, Global Memory GM, Quiesced ESV Buffer BQ, and Loop Buffers, B1, B2, B3, B4.

Program Memory PM serves for storing user and system programs. Data Memory DM is a one-level storage device for data, and a work storage. Global Memory GM is similar in design to Data Memory, but is intended for the purpose of setting and resetting of switches (semaphores) for global use by all tasks in progress. Buffers B1 and B2 serve as synchronization buffers at the interface of their loops, IFL and BL, with the Feeder Loop FL. They are dribble-down type buffers. Buffers B3 and B4 solely increase the storage capacity of the Feeder Loop and the Main Loop, respectively. They are static buffers of constant size. The Quiesced ESV Buffer BQ serves for storage of external state vectors temporarily quiesced (i.e, in wait state). Ei...