Browse Prior Art Database

Conditional Cache Miss Facility for Handling Short/Long Cache Requests

IP.com Disclosure Number: IPCOM000049479D
Original Publication Date: 1982-Jun-01
Included in the Prior Art Database: 2005-Feb-09
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Losq, JJ: AUTHOR [+4]

Abstract

The degradation that occurs in a cache-managed scalar/vector machine when long vector requests encounter multiple cache misses is eliminated, by bypassing the cache and sending data directly to the processor on vector operations of greater than one line. This equals or improves upon cache access and avoids purging of the cache by moving long vectors through the cache.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 56% of the total text.

Page 1 of 1

Conditional Cache Miss Facility for Handling Short/Long Cache Requests

The degradation that occurs in a cache-managed scalar/vector machine when long vector requests encounter multiple cache misses is eliminated, by bypassing the cache and sending data directly to the processor on vector operations of greater than one line. This equals or improves upon cache access and avoids purging of the cache by moving long vectors through the cache.

A cache-managed scalar/vector machine intends to service requests for scalar operands as well as for short/long vector ones. The cache requests that encounter cache misses are handled by the Cache Miss Facility (CMF). This facility manages the cache data line traffic between the cache, main memory and the processor.

The mode of operation described pertains to a Conditional Cache Miss Facility (CCMF), whereby the cache requests are handled depending on the type of the requested access. For anticipated multiple cache misses, the cache and CCMF are virtually bypassed, since the line transfer is effectuated directly from memory (DMA- Direct Memory Access). Such handling of the CCMF can result in performance enhancement, especially for long vector requests.

For scalar and vector requests the CCMF examines the cache miss in order to determine the type of the request. For an encountered cache miss of a scalar and/or short vector request, the CCMF dispatches the line's fetch address to the main memory unit in order to initiate the line t...