Performance visualization of parallel programs
Performance visualization of parallel programs
Performance is a critical issue in current massively parallel processors. However, delivery of adequate performance is not automatic and performance evaluation tools are required in order to help the programmer to understand the behaviour of a parallel program. In recent years, a wide variety of tools have been developed for this purpose including tools for monitoring and evaluating performance and visualization tools. However, these tools do not provide an abstract representation of performance. Massively parallel processors can generate a huge amount of performance data and sophisticated methods for representing and displaying this data (e.g. visual and aural) are required. Current performance views are not scalable in general and do not represent an abstraction of the performance data.
The Do-Loop-Surface display is proposed as an abstract representation of the performance of a particular do-loop in a program. It has been used to improve the performance of several algorithms on different hardware platforms (e.g. CM-5, Meiko CS-2, IBM SP2). The examples demonstrate that the Do-Loop-Surface display is an useful way to represent performance. It is implemented using AVS (Application Visualization System), a standard data visualization package. The use of scientific visualization tools such as AVS to display performance data, is becoming a very powerful alternative to support performance analysis of parallel programs. The Do-Loop-Surface (DLS) display is presented in this thesis as an example on how a data visualization tool can be used to define new abstract representations of performance, helping the user to analyze complex data potentially generated by a large number of processors.
Additionally to this main contribution, our experience developing the performance tool ANDES is presented in this thesis as well as our studies related to the invasiveness effects that performance instrumentation can generate.
D'Paola, Oscar Naim
01e5240e-6c46-4eb5-a91a-23146b469e3f
May 1995
D'Paola, Oscar Naim
01e5240e-6c46-4eb5-a91a-23146b469e3f
Hey, A.J.G.
a3cd725f-374f-488f-b1cf-ba671cfa6a7d
D'Paola, Oscar Naim
(1995)
Performance visualization of parallel programs.
University of Southampton, Electronics & Computer Science, Doctoral Thesis, 145pp.
Record type:
Thesis
(Doctoral)
Abstract
Performance is a critical issue in current massively parallel processors. However, delivery of adequate performance is not automatic and performance evaluation tools are required in order to help the programmer to understand the behaviour of a parallel program. In recent years, a wide variety of tools have been developed for this purpose including tools for monitoring and evaluating performance and visualization tools. However, these tools do not provide an abstract representation of performance. Massively parallel processors can generate a huge amount of performance data and sophisticated methods for representing and displaying this data (e.g. visual and aural) are required. Current performance views are not scalable in general and do not represent an abstraction of the performance data.
The Do-Loop-Surface display is proposed as an abstract representation of the performance of a particular do-loop in a program. It has been used to improve the performance of several algorithms on different hardware platforms (e.g. CM-5, Meiko CS-2, IBM SP2). The examples demonstrate that the Do-Loop-Surface display is an useful way to represent performance. It is implemented using AVS (Application Visualization System), a standard data visualization package. The use of scientific visualization tools such as AVS to display performance data, is becoming a very powerful alternative to support performance analysis of parallel programs. The Do-Loop-Surface (DLS) display is presented in this thesis as an example on how a data visualization tool can be used to define new abstract representations of performance, helping the user to analyze complex data potentially generated by a large number of processors.
Additionally to this main contribution, our experience developing the performance tool ANDES is presented in this thesis as well as our studies related to the invasiveness effects that performance instrumentation can generate.
Text
95088358.pdf
- Other
More information
Published date: May 1995
Organisations:
University of Southampton, Electronics & Computer Science
Identifiers
Local EPrints ID: 365532
URI: http://eprints.soton.ac.uk/id/eprint/365532
PURE UUID: 20337e82-2105-449b-958a-8f90150fb5d7
Catalogue record
Date deposited: 09 Jun 2014 11:51
Last modified: 14 Mar 2024 16:56
Export record
Contributors
Author:
Oscar Naim D'Paola
Thesis advisor:
A.J.G. Hey
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics