Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Automated Analysis of Memory Leakage using Linear Regression Modeling

IP.com Disclosure Number: IPCOM000116644D
Original Publication Date: 1995-Oct-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 4 page(s) / 72K

Publishing Venue

IBM

Related People

Corn, CF: AUTHOR [+2]

Abstract

A method for automating Memory Leak analysis is disclosed. The measured leakage data points are used to develop a Simple Linear Regression Model, which provides quick and reliable analysis for leakage which is fairly linear.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 53% of the total text.

Automated Analysis of Memory Leakage using Linear Regression Modeling

      A method for automating Memory Leak analysis is disclosed.  The
measured leakage data points are used to develop a Simple Linear
Regression Model, which provides quick and reliable analysis for
leakage which is fairly linear.

      The best way to describe how this invention works, is to
demonstrate using an example.  On the Beta release of a product,
Memory Leak Data was collected for each process.  Since the code was
relatively stable, data was collected for a long period of time (74
hours).  For the follow on development phase of the product, Memory
Leak Data was also collected for each process.  Since the code was
relatively unstable, data could only be collected for a short period
of time (17 hours).  The Diagrams A (Beta) and B (Post Beta)
illustrate the Memory Leak data points graphed for the two different
versions of the product.

      In these examples, it was apparent that the leakage was
somewhat linear.  By using the measured data points and applying the
formulas for developing a Simple Linear Regression Model, the linear
leakage model for each of the different processes was developed:

   Process Memory Leakage =  intercept  +  slope (Time)

The following summary output illustrates the automated analysis using
the calculated slope generated for the two different versions of the
product:
  Modeling Data for Diagram A    |  Modeling Data for Diagram B
  Product:           Beta        |     Produ...