Browse Prior Art Database

Method and System for a Simplified Restore of a Mapper State after a Flush Operation

IP.com Disclosure Number: IPCOM000198649D
Publication Date: 2010-Aug-11
Document File: 2 page(s) / 47K

Publishing Venue

The IP.com Prior Art Database

Abstract

A method and system for a simplified restore of a mapper state after a flush operation is disclosed. In this method, a bit is set which indicates that until a write for the indicated register resource occurs, all valid earlier issue queue operations may be used.

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 54% of the total text.

Page 1 of 2

Method and System for a Simplified Restore of a Mapper State after a Flush Operation

Disclosed is a method and system for a simplified restore of a mapper state after a flush operation.

Typically, in an out-of-order system, issue queue (QPOS) valid tracking information must be maintained for all writes to a resource that have not yet completed. This tracking is required only to restore QPOS tracking information on a flush. This can represent a lot of area and complexity for resources that are written often, such as the condition code (CC.)

The method and system disclosed herein provides a way to track only the most recent QPOS valid, and on a flush readers are made dependent on all valid older issue queue entries until a write for the indicated register resource occurs.

The figure illustrates a condition code mapper using the disclosed mapper scheme. The condition code mapper illustrates the last write information, a RTAG backup table and a QPOS valid update logic. Whenever a new writer comes in, the control logic provides a new RTAG to use for the new write operation, the issue queue provides a new QPOS/valid for setting the dependency information. This new information is then stored into the last write register. Whenever a new writer comes in, the depend

_

                                                   all bit is always set to zero. During normal operation, whenever the issue queue retires an entry, the information is passed to the QPOS valid logic that checks if the information is the QPOS ID associated with the last write. If so, the QPOS valid logic clears the
qpos

_valid bit. If a new write comes in and re...