Browse Prior Art Database

Method and Apparatus for Monitoring CPU Loading in a Multiprocessor Environment

IP.com Disclosure Number: IPCOM000009986D
Original Publication Date: 2002-Oct-03
Included in the Prior Art Database: 2002-Oct-03
Document File: 2 page(s) / 28K

Publishing Venue

Motorola

Related People

Todd Schaefer: AUTHOR [+2]

Abstract

Embedded systems may make use of a host processor that is responsible for distributing workloads across multiple slave processors. An evenly distributed workload across the multiple slave processors is the most efficient use of resources. To evenly distribute the workload the host processor requires knowledge of the current level of CPU loading in the slave processors. This is accomplished through a combination of software and hardware to perform real-time sampling of the slave processors. This CPU loading information is available to the host processor so that it can make real-time workload distribution decisions.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 50% of the total text.

Method and Apparatus for Monitoring CPU Loading in a Multiprocessor Environment[MP1] 

By Todd Schaefer, Paul Kos

Motorola, Inc.

Global Telecom Solutions Sector

 

ABSTRACT

Embedded systems may make use of a host processor that is responsible for distributing workloads across multiple slave processors. An evenly distributed workload across the multiple slave processors is the most efficient use of resources.

To evenly distribute the workload the host processor requires knowledge of the current level of CPU loading in the slave processors. This is accomplished through a combination of software and hardware to perform real-time sampling of the slave processors. This CPU loading information is available to the host processor so that it can make real-time workload distribution decisions.

PROBLEM

A 3G cellular network card converts ATM traffic to Ethernet traffic, and vice versa. The network card has a host processor with several dedicated slave processors on which call traffic is handled. The host processor is responsible for optimally balancing calls across all slave processors. The slave processors are dedicated processors responsible for the ATM/IP conversions.

Calls cannot simply be assigned in a round-robin fashion or randomly across the dedicated slave processors because various factors can influence CPU loading such as: calls have different data rates, different size packet payloads, slower/faster memory devices, shared-bus interaction, voice/data calls, etc.

Due to influential factors affecting CPU loading and the dynamic nature of call setup/teardowns and call length there needs to be a real-time method of monitoring CPU loading on the dedicated slave processors.

SOLUTION

The solution for real-time monitoring of the sla...