Browse Prior Art Database

Resource governor optimization based on application workload forecast

IP.com Disclosure Number: IPCOM000238680D
Publication Date: 2014-Sep-11
Document File: 1 page(s) / 21K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a technology to use software application to forecast/estimate the workload and adjust the processor frequency based on it.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 99% of the total text.

Page 01 of 1

Resource governor optimization based on application workload forecast

Introduction:

There are various computing resource governor existing today, e.g., P-state governor in the OS that changes the CPU frequency based on the current workload. However, it's very hard for OS or hardware to predict or foresee the workload. Currently the governors can only adjust the resource utilization level based on the instant workload. For example, sometimes the CPU frequency are set to a higher than necessary rate just due to a short-time spike of workload.

The idea:

Let the software application do the workload forecast/estimation. The application, based on the current task it has, reports a "progress bar" information to the governor. The governor then works based on the forecast information and adjusts the resource utilization. For example, adjust the processor frequency based on the remaining workload estimated by the applications.

The details:


1. The application notify the resource governor that it is going to estimate the workload.

2. The governor start to record the CPU frequency and other resources (like I/O access) utilization used by the application.

3. The application finishes the workload estimation and reports the remaining percentage of workload to the governor.

4. The governor, based on the information collected, gets a better understanding what resources are going to be required by the application in the near future.

5. The governor adjust the resource management p...