Browse Prior Art Database

Range-Driven Prefetching of Cache Lines

IP.com Disclosure Number: IPCOM000041688D
Original Publication Date: 1984-Mar-01
Included in the Prior Art Database: 2005-Feb-02
Document File: 1 page(s) / 11K

Publishing Venue

IBM

Related People

Hoevel, LW: AUTHOR [+2]

Abstract

It has been established that misses in a cache memory are clustered in so-called bursts. These bursts are the results of transitions between operating system modules (changes of locality). It has also been found that most of the bursts are made up of a small range of consecutive lines (unit of paging in the cache). The number of misses in the cache and the finite cache penalty are reduced by the following algorithm: at the first miss in the range, prefetch the remaining lines in the range. It is proposed that the table be initialized by a software module, triggered by an operator command.

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

Page 1 of 1

Range-Driven Prefetching of Cache Lines

It has been established that misses in a cache memory are clustered in so- called bursts. These bursts are the results of transitions between operating system modules (changes of locality). It has also been found that most of the bursts are made up of a small range of consecutive lines (unit of paging in the cache). The number of misses in the cache and the finite cache penalty are reduced by the following algorithm: at the first miss in the range, prefetch the remaining lines in the range. It is proposed that the table be initialized by a software module, triggered by an operator command.

1