Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Parallel computer systems
Koskela R., Simmons M., ACM Press, New York, NY, 1990. Type: Book (9780201509373)
Date Reviewed: May 1 1992

Papers and summaries from the Workshop on Parallel Computer Systems: Performance Instrumentation and Visualization, held in May of 1989 in Santa Fe, New Mexico, are collected here. Most of the chapters describe current research in this area; a collection such as this allows comparisons between different projects. This book is neither a tutorial nor appropriate as the sole source of material for a course; it can serve as a reference for workers interested in this field.

Several chapters show a trend toward design and use of hardware performance monitors. Such monitors are described for the INCAS distributed multicomputer at the University of Kaiserslautern, for the iPSC hypercube at Illinois, and for the IBM RP3. Two chapters describe performance instrumentation designed into commercial systems. The Gould NP1 reflective memory was designed to support shared memory communications at high speed over relatively long distances (35 meters) and was instrumented to signal events or watch resource utilization. The Sequent multiprocessors were instrumented to watch cache coherence effectiveness; the chapter discussing them also shows extensive results of instrumented code running on that commercial system. No consensus is reached in this book, but my reading is that many of the authors believe that hardware performance monitors will become standard for parallel systems at some point in the future.

Several other chapters show designs for performance visualization, such as for the IBM RP3 and for parallel LISP, as well as other event and performance display systems. Novel techniques to display large quantities of performance data are required to avoid overloading the user with too much data and too little information. The work shown in this book does not convince me that it will scale the performance display up to large numbers of processors. The book also includes short reports from four working groups, one of which tries to define a standard interface from a programming language to a performance instrumentation system, and standard formats for trace files.

The book suffers from a little disorganization and a few chapters only loosely coupled to the main subject, but as a concise snapshot of the current state of performance instrumentation and visualization, it belongs on the shelf of anyone working in or interested in this area. For those interested in starting research projects along this line, this book will serve as a reference and source of ideas.

Reviewer:  M. Wolfe Review #: CR115140
Bookmark and Share
 
Parallel Programming (D.1.3 ... )
 
 
Parallel Processors (C.1.2 ... )
 
 
Performance Attributes (C.4 ... )
 
 
Performance Measures (D.2.8 ... )
 
 
Super (Very Large) Computers (C.5.1 ... )
 
 
Performance (D.4.8 )
 
Would you recommend this review?
yes
no
Other reviews under "Parallel Programming": Date
How to write parallel programs: a first course
Carriero N. (ed), Gelernter D. (ed), MIT Press, Cambridge, MA, 1990. Type: Book (9780262031714)
Jul 1 1992
Parallel functional languages and compilers
Szymanski B. (ed), ACM Press, New York, NY, 1991. Type: Book (9780201522433)
Sep 1 1993
Data-parallel programming on MIMD computers
Hatcher P., Quinn M., MIT Press, Cambridge, MA, 1992. Type: Book (9780262082051)
Sep 1 1993
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy