eugene@pioneer.arpa (Eugene Miya N.) (08/18/87)
This is a short review of two books I had a chance to read while riding
to Lake Tahoe over the weekend. The reviews represent my opinions.
%A G. Jack Lipovski
%A Miroslaw Malek
%T Parallel Computing: Theory and Comparisons
%I Wiley-Interscience
%C New York
%D 1987
%K computational energy and efficiency,
%X The long awaited book on the Texas Reconfigurable Array.
It is a bit disappointing (not even a photograph), and the diagrams are not
completely clear. The meat of the subject does not enter until late.
The concept of inductive computing is interesting. Very little about
TRACOS, CSL, or other software related issues.
The appendices are a collection of previously published papers:
PASM, Ultracomputer, IBM RP3, Cedar, Cosmic Cube, CHiP, non-Von.
There is a very nice paper on the problem of scaling. The majority of
the books is about interconnection networks (not parallel computing).
The annotation which I put into my bibliography says most of what I
think about this book. It is largely disappointing. Buy this book
for the institutional library, but I note a special paper from MCC
down below. I have written for copies of papers from this project, but
have never received any. Very little empirical data from their work.
There are improperly cited references (see Page 170 for Kahn which never
appears in the Reference section).
I expect I will get some criticism from someone from Texas after reading
this review.
%A Edward Gehringer
%A Daniel P. Siewiorek
%A Zary Segall
%Z CMU
%T Parallel Processing: The Cm* Experience
%I Digital Press
%C Boston, MA
%D 1987
%K Book,
%X Looks okay!
I have been accused of a slight CMU bias. In my defense, I have talked
to most of the people who have worked on various CMU projects: C.mmp,
Cm*, C.vmp, and their successors. They have freely given out
information, and I note that all the PhD students have gotten out of
parallel processing beause of the difficulties of the problem. There is
loads of quantitiative data. They have worked very hard on some
difficult problems. Some of this book consists of modification to
technical reports and PhD theses, so some fault might be found here.
These people have been accused of poor designs [working in the confines of DEC
products], having poorly programmable systems [by people who have not
programmed parallel algorithms themselves and don't know the problems or
difficulties], but they did build several machines and get useful
empirical data [to criticism from those with paper designs].
Note: the appendices of the L&M book considers other architectures (many
not yet built) in a `critical' light [critical commentay lasts 1-3
pages]. The Cm* book elaborates on the `experiments' done on Cm*.
A reader can fault the lack of comparison with other systems, but they
can't fault real data, much of which will probably be rediscovered at a
later date by some researcher from a makers of big machines.
While this is a real old project, readers should note that TRAC and Cm*
started roughly the same time [their original papers published rough the
same time late 1977/78]. These books also come out about the same
time. Their comparison offers an interest insight as to the length of
time it takes to embark on major parallel processing projects during
a time when the subject is not popular, when languages and operating
systems are not standardized, and when people have a hard time realizing
that we really don't know how to build efficient parallel machines.
Back to the TRAC book. There is one special paper which is reproduced
in Lipovski and Malek which is something of a gem. There is a cute
analogy to monster movies. Don't ask me for a copy, either ask the
authors or buy the L&M book.
%A Alfred C. Hartman
%A Jeffery D. Ullman
%T Model Categories for Theories of Parallel Systems
%R TR PP-341-86
%I MCC
%C Austin, TX
%D 1986
%X A nice little technical paper on complexity (starting with VLSI)
and scale.
Reproduced in "Parallel Computing: Theory and Comparisons,"
by G. Jack Lipovski and Miroslaw Malek,
Wiley-Interscience, New York, 1987, pp. 369-381, appendix H.
From the Rock of Ages Home for Retired Hackers:
--eugene miya
NASA Ames Research Center
eugene@ames-aurora.ARPA
"You trust the `reply' command with all those different mailers out there?"
"Send mail, avoid follow-ups. If enough, I'll summarize."
{hplabs,hao,ihnp4,decwrl,allegra,tektronix,menlo70}!ames!aurora!eugene