Browse Prior Art Database

Diagnostic Pre-Run Queueing

IP.com Disclosure Number: IPCOM000113512D
Original Publication Date: 1994-Aug-01
Included in the Prior Art Database: 2005-Mar-27
Document File: 2 page(s) / 31K

Publishing Venue

IBM

Related People

Ash, KJ: AUTHOR [+2]

Abstract

A method for improving performance of diagnostic execution on a partitioned resource in a multiprocessing environment is disclosed. Diagnostics are queued concurrently on all available processors, removing restrictions designed to improve performance.

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

Diagnostic Pre-Run Queueing

      A method for improving performance of diagnostic execution on a
partitioned resource in a multiprocessing environment is disclosed.
Diagnostics are queued concurrently on all available processors,
removing restrictions designed to improve performance.

      It is desirable to run diagnostics on resources before
releasing them to a computer system.  On large resources, such as
memory, it is possible to partition the resource into regions to run
diagnostics.

      Hardware designs may enable diagnostics to correct for errors.
The correction of these defects, and subsequent checking, takes a
significant amount of time.  However, it is important to bring the
resource on-line as quickly as possible, to permit system operation.

      Uniprocessing systems require a serial approach, checking each
region in turn.  Speeding up initial bring-up is accomplished by
deferring bring-up on regions that have potentially repairable
defects.

      In a multiprocessing environment, it is possible to run the
diagnostics in parallel on multiple regions.  When the ratio of
processors to resources is sufficiently high, it is possible to
eliminate the delaying of defect correction.  This improves the time
to bring the entire resource on-line, by eliminating the re-detection
of the initial error which forced execution deferral.