Browse Prior Art Database

Mechanism for Cache Replacement and Prefetching Driven by Heuristic Estimation of Operating System Behavior

IP.com Disclosure Number: IPCOM000051481D
Original Publication Date: 1981-Jan-01
Included in the Prior Art Database: 2005-Feb-10
Document File: 1 page(s) / 11K

Publishing Venue

IBM

Related People

Hoevel, LW: AUTHOR [+2]

Abstract

The following mechanism reduces finite cache penalty and improves efficiency of cache hardware in environments where CSETS (modules) of the MVS (Multiple Virtual Storage) operating system are called in a cyclic pattern. In such environment, given that MVS is executing a particular module, we can predict with confidence which module is likely to be called next.

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

Page 1 of 1

Mechanism for Cache Replacement and Prefetching Driven by Heuristic Estimation of Operating System Behavior

The following mechanism reduces finite cache penalty and improves efficiency of cache hardware in environments where CSETS (modules) of the MVS (Multiple Virtual Storage) operating system are called in a cyclic pattern. In such environment, given that MVS is executing a particular module, we can predict with confidence which module is likely to be called next.

It is proposed that a table of module addresses (beginning address and end address of the module) be built after IPL (initial program loading); this table will then be used by the least recently used mechanism for cache replacement to insure that the replacement candidate is a line which has not been recently used and which is not likely to be used in the near future. Our experiments show that static transfer statistics are sufficient to predict which area (module) is likely to be used in the near future based only on the knowledge of which module is currently active.

Additional benefits can be obtained by permitting a line that has just been used to be a candidate for replacement if we know, because of the table, it is very unlikely to be used in the near future.

Prefetching of cache lines based on this table should also be more efficient because it will then be known what is going to be called next.

1