Browse Prior Art Database

Data Structure and Insertion Algorithm for Representing Asynchronous Occurrences for Visualization by Trace Visualization Tools using Ghant Charts with Occurrence Hierarchies

IP.com Disclosure Number: IPCOM000105426D
Original Publication Date: 1993-Jul-01
Included in the Prior Art Database: 2005-Mar-19

Publishing Venue

IBM

Related People

Lehr, TF: AUTHOR

Abstract

Disclosed are a data structure and insertion algorithm for visualizing asynchronous occurrences in event-driven traces by visualization tools using Ghant charts with occurrence hierarchies. An example of an event-driven trace facility is the AIX* 3.2 trace Facility [1]. An event is an instant recorded by the execution of a hook, and an occurrence is a pair of related events (the entry and exit to a procedure call, for example). Ghant charts are data visualization charts. A simple Ghant chart is better known as a bar chart or histogram in which several rectangles of varying length, representing different quantities, are arranged parallel to some axis. Simple Ghant charts are often used to represent counts or statistical quantities.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 20% of the total text.

Data Structure and Insertion Algorithm for Representing Asynchronous Occurrences for Visualization by Trace Visualization Tools using Ghant Charts with Occurrence Hierarchies

      Disclosed are a data structure and insertion algorithm for
visualizing asynchronous occurrences in event-driven traces by
visualization tools using Ghant charts with occurrence hierarchies.
An example of an event-driven trace facility is the AIX* 3.2 trace
Facility [1].  An event is an instant recorded by the execution of a
hook, and an occurrence is a pair of related events (the entry and
exit to a procedure call, for example).  Ghant charts are data
visualization charts.  A simple Ghant chart is better known as a bar
chart or histogram in which several rectangles of varying length,
representing different quantities, are arranged parallel to some
axis.  Simple Ghant charts are often used to represent counts or
statistical quantities.

      More complex Ghant charts use rectangles to represent elapsed
time of occurrences like procedures and system calls.  If the purpose
of the Ghant chart is to chronicle the occurrences recorded by a
trace, the rectangles may lie end-to-end in order to indicate the
execution se quence of the occurrences.  Berstein, Bolmarcich and So
[2]  use primitive Ghant charts very effectively in order to view
parallelism of Fortran applications.  They, however, can not easily
increase the number of kinds of occurrences they wish to visualize
because they do not permit occurrence hierarchies.  Sevcik [3]  uses
more sophisticated Ghant charts to visualize parallelism profiles.
He uses a simple notion of occurrence hierarchies in representing the
parallel occurrences but provides no general way of providing an
unlimited hierarchy.  Sophisticated trace visualization Ghant charts
like Piescope of the Parallel Instrumentation Environment (PIE) [4]
permit occurrence hierarchies where rectangles may lie on top of one
another to indicate that one occurrence executed within the context
of another, superior occurrence.  For example, if the AIX 3.2 trace
facility records that a read() system call makes a call to a socket
operation, the system call is a superior occurrence with respect to
the socket operation.  A Ghant chart supporting occurrence
hierarchies would display one rectangle representing the execution of
time of the socket operation lying within a larger rectangle
representing the execution time of the system call.

      An example of a Ghant chart with occurrence hierarchies is
shown in Fig. 1.  Fig. 1 is a Piescope display depicting part of an
execution trace of the invocation of an aixterm window.  Processes
are numbered and named on the vertical axis.  Time, measured in
seconds, is on the horizontal axis.  The different shaded rectangles
represent occurrences like system calls, socket operations and
processes.  Examine process 6719, the X server.  One sees several
rectangles lying end-to-end parallel to the...