Browse Prior Art Database

X Windows Application Decomposition and Performance Analysis Tool

IP.com Disclosure Number: IPCOM000112448D
Original Publication Date: 1994-May-01
Included in the Prior Art Database: 2005-Mar-27
Document File: 2 page(s) / 30K

Publishing Venue

IBM

Related People

Murrell, DB: AUTHOR [+2]

Abstract

A program is disclosed that decomposes X applications into X protocols, allowing performance analysis at the protocol level. An analyst or developer can determine which protocol uses the most time, and tune that protocol. Given that tool difficulty of use makes tools less likely to be used, this tool includes a Graphical User Interface that increases its ease of use.

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

X Windows Application Decomposition and Performance Analysis Tool

      A program is disclosed that decomposes X applications into X
protocols, allowing performance analysis at the protocol level.  An
analyst or developer can determine which protocol uses the most time,
and tune that protocol.  Given that tool difficulty of use makes
tools less likely to be used, this tool includes a Graphical User
Interface that increases its ease of use.

      The first step in analysis is to use the tool, known as PIE, to
record a key sequence of an X application, using Xrecorder.  If a
recording already exists, it can be used.

      Next, run the recording through PIE and obtain a timing.  PIE
uses trace hooks imbedded in the X server to perform the timing.  It
keeps track of how long each protocol request takes, adding like
protocol requests together.  When the application decomposition is
complete, the developer has a table showing which protocols the
application called and how much time each used.

      Using the timing information, analysts and developers tune the
X server and adapter device dependent code to decrease the time
spent.

      After this tuning, PIE is rerun, new protocols are targeted and
tuned.  These steps are iterated until performance is acceptable.