Browse Prior Art Database

Self Correcting Duplex Memory

IP.com Disclosure Number: IPCOM000096864D
Original Publication Date: 1963-Dec-01
Included in the Prior Art Database: 2005-Mar-07
Document File: 2 page(s) / 48K

Publishing Venue

IBM

Related People

McNeil, CV: AUTHOR

Abstract

Where stringent reliability requirements exist for a computer memory, one solution is to apply redundancy techniques. This memory system consists basically of two identical memory subsystems with relating and checking apparatus for providing readout from only the correctly operating subsystem.

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 86% of the total text.

Page 1 of 2

Self Correcting Duplex Memory

Where stringent reliability requirements exist for a computer memory, one solution is to apply redundancy techniques. This memory system consists basically of two identical memory subsystems with relating and checking apparatus for providing readout from only the correctly operating subsystem.

Each subsystem includes a memory module, a buffer register and a parity check, designated as triple modular redundant. Memory selection logic, also triple modular redundant, receives error detector output signals from each of the modules as well as corresponding signal information from the registers and parity check. peripheral instrumentation, not shown, is also provided for each of the memory modules such as control address drivers, inhibit drivers, and the like.

Assuming first that memory A is being used and that an error is detected via parity check, operation immediately ceases in the A memory readout. Operation transfers to the B memory readout through the operation of the memory select logic to memory A inhibit drivers via connection logic A and I. Both memories are at this time regenerated by the buffer register of the good memory, namely, memory module B, thus correcting for any transient errors.

After a second parity check and error detection verification that a memory error no longer exists, each memory is again returned to the control of its own buffer register. However, operation is maintained. in the B memory until it experiences...