Browse Prior Art Database

Managing Targets of Multiprocessor Cross Interrogates

IP.com Disclosure Number: IPCOM000045970D
Original Publication Date: 1983-May-01
Included in the Prior Art Database: 2005-Feb-07
Document File: 1 page(s) / 11K

Publishing Venue

IBM

Related People

Puzak, TR: AUTHOR [+3]

Abstract

Cross-interrogate actions on multiprocessor systems can cause performance degradation. The following relieves the problem by using least recently used (LRU) mechanisms.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 100% of the total text.

Page 1 of 1

Managing Targets of Multiprocessor Cross Interrogates

Cross-interrogate actions on multiprocessor systems can cause performance degradation. The following relieves the problem by using least recently used (LRU) mechanisms.

The handling of lines which are subject of cross-interrogate action can cause multiprocessor degradation if the lines "ping-pong" between the caches or cause access delay if the ping-pong is avoided by use of a shared cache (for cross- interrogated lines). Analysis of LRU status of lines subject to cross interrogate show that they are the most recently used (MRU) within their congruence class and that such lines account for the majority of all subsequent additional cache misses associated with multiprocessor (MP) invalidations and cache cast-outs. As such the LRU status forms a basis for handling cross-- interrogated lines either as the candidates for shared cache (MRU) or directly transferred over to the requester's cache if they are not MRU.

1