The authors interviewed people working for 55 academic, financial, and technical organizations in Europe (3), the United States (30), and Japan (22). Their primary message is that competitiveness demands a new kind of thinking; they strongly suggest that we have to move into the future, where that means the mastery of artificial intelligence, the use of powerful parallel systems, and learning to think in parallel. The book is intended, according to the jacket, for “farsighted CIOs, corporate managers, and investment professionals.”
The first third of the book (chapters 1 through 4) is motivational and lays out a scenario supporting the primary themes of parallelism, artificial intelligence, and increasing computational power. Chapter 5 is called “From New Components to Hypercubes.” Chapter 6 covers the Connection Machine. Chapter 7 describes the iPCS/2 and chapter 8 is on the BBN Butterfly. Chapter 9 discusses “Array Processors of Cray Research” and chapter 10 covers the ETA Cryogenic Computer. Chapters 11 and 12 discuss the Japanese new-generation computer project. Chapters 13 through 15 cover “Databases for Ideas and Documents,” the Teradata Database Engine, and database applications. An epilogue discusses the authors' view of sixth-generation technology. The index is adequate. References appear in the text and in footnotes, but they are few. The material is mostly textual, with little use of figures.
This book is fraught with problems, beginning with the title; a more appropriate title would be Advanced information systems. It is not a text about the most advanced computers, as the title suggests, although part of the material is devoted to high-end machines. The writing style often borders on hyperbole. For example, the authors make the claims that “the fact is increasingly accepted that prior to 1995 artificial intelligence and classical data processing would have merged, with the former being the surviving discipline in the design and application of information systems” and “current projections talk of photonics replacing electronics by the end of this century.” I see little, if any, possibility of these events occurring in these time frames.
In other cases the text is either wrong, misleading, or so poorly written as to be either unclear or incomprehensible. For example, on MIMD, SIMD, and multiple SIMD, the authors say, “they will not be held back because of programmer resistance to new machines that do not port existing Fortran-based programs.” Their comment on workstations is, “Since the advent of the personal computer and the dedicated workstation (WS) was made feasible at an affordable level, a valid information systems solution starts and ends at the WS level—although it may involve a variety of central and departmental resources.” About the difficulty of parallel programming, they say, “Another scope of parallelism is simplicity of programming. Contrary to von Neumann-type machines, 5 GCs are looking at the whole problem at once. This is a more natural approach to computing. It results in simpler programming because there is no need to break up a program, assign its parts to different jobs, and then interconnect them, eventually developing very large software structures.” In the chapter on Cray computers, the authors indicate that Cray systems were the first widely accepted array processors, which is not the usual definition of the Cray architecture. In many cases these errors are only minor distortions of accepted systems descriptions, but they are found throughout the text and add up to a significantly distorted view of the technologies described.
While I agree with the main message the authors are trying to convey, namely that we need to rethink many of our precepts of computing and boldly step into the future, the supporting arguments and clear action plans are missing. The reader is left with the impression that the authors want us to take the bold step forward, but they have not justified it in a sensible way or provided any useful guidance for implementation. At best, they have misinterpreted the interview data and present a superficial view of the future.
I do not recommend this book. It overstates the case for advanced technology, is not well organized or technically sound, and does not adequately justify the conclusions reached. The chapters on specific systems are little more than overviews and do not contribute to a better understanding of the systems described.
Access critical reviews of Computing literature here
IPDPS'06: Proceedings of the 20th international conference on Parallel and distributed processing
The HPC Challenge (HPCC) benchmark suite and the Intel MPI Benchmark (IMB) are used to compare and evaluate the combined performance of processor, memory subsystem and interconnect fabric of five leading supercomputers - SGI Altix BX2, Cray X1, Cray ...
The HPC Challenge (HPCC) Benchmark suite and the Intel MPI Benchmark (IMB) are used to compare and evaluate the combined performance of processor, memory subsystem and interconnect fabric of five leading supercomputers-SGI Altix BX2, Cray X1, Cray ...
IPDPS '12: Proceedings of the 2012 IEEE 26th International Parallel and Distributed Processing Symposium
Gemini, the network for the new Cray XE/XK systems, features low latency, high bandwidth and strong scalability. Its hardware support for remote direct memory access enables efficient implementation of the global address space programming languages. ...