Паттерсон компьютерная архитектура количественный подход руководство

Computer Architecture: A Quantitative Approach

Authors: Hennesy & Patterson
Price: $71.35 at FatBrain; $83.95 at Amazon
Target Audience: EE students and those with a serious thirst for CPU knowledge. This is manna for CPU freaks, but overkill for anyone who doesn’t wanna bury their mind in it.

Rating: 5 out of 5 woo-hoos!

People often ask me for my opinion on the best way to gain a good understanding of modern CPU architecture. My reply is invariably that you should read, read, read—read everything you can get your hands on, whether you think you’ll understand it or not. As students of anything esoteric already know, if you read enough stuff that you barely understand then pretty soon you’ll start to make connections and things will begin to fall together. Terms will make sense, and you’ll get progressively more enlightened until at some point you’ll be In The Know. The next obvious question is, what should one read? While there are a number of good technical resources available on the Web, the best place to go for a real education is still books. And I don’t mean, “Computer Architecture for Dummies” or “Learn Computer Architecture in 10 Days.” If you really want to try to wrap your mind around concepts like pipelining, vector processing, ISA design, RISC, CISC, and all the other terms you hear tossed about, you’ve got to tackle the big stuff.

Computer Architecture: A Quantitative Approach, is the big stuff. Odds are that if you were to take a computer architecture class at almost any university, this would be the book you’d use. (The only exception to this would be if you took a class with a professor who has written a competing book on computer architecture. Then you’d use his book.) With the first edition published in 1990 and the second edition published in 1996, Computer Architecture is relatively up to date, and it’s what the chip designers of tomorrow are cutting their teeth on.

If anyone is qualified to talk about computer architecture, it’s David A. Patterson and John L. Hennessy. Patterson was at the vanguard of the RISC movement; he published a number of seminal RISC papers and even coined the term “RISC.” He was the man behind the design of Berkely’s RISC-I computer, the machine on which Sun’s SPARC was based. His other contributions include leadership of the team that invented RAID, as well as research into computation using networks of workstations (NOW). John Hennessy was also one of the leaders of the RISC movement in the early 80’s. He worked on the Stanford team that developed the MIPS architecture, and he later went on to found MIPS Computer Systems, which eventually merged with SGI. If you’re a member of the ACM and/or you have access to their Digital Library, I’d recommend doing a search for either of these authors and reading some of their other published work.

Fortunately for computer architecture students,  Hennessy and Patterson are excellent writers. Often the lectures in senior-level CS and EE classes are awful; if you had a truly gifted lecturer for a computer architecture class then you were one of the fortunate few. But since you’re probably using Hennessy and Patterson (this is how Computer Architecture is referred to by computer architects), the quality of the lectures is of little importance. If you’ve got a good enough technical background, you can grasp most of the concepts in it just by reading the book thoroughly. Computer Architecture is by no means easy though, but its difficulty lies not in the presentation but in the subject matter. Because of the advanced nature of the subject matter, it’s not really a book for beginners. Nevertheless, if you’re willing to invest some serious effort and time into learning about modern microprocessor design, then even those with only basic programming and technical skills can get a lot out if it. (I got email once from a reader who’s a music major and who claims he read the book to great benefit.) If you don’t have a solid technical background then you shouldn’t expect to be able to grasp everything in the book, but you can still increase your knowledge immensely.

The book contains chapters on basic computer design, instruction set principles, pipelining, instruction-level parallelism, networks, multiprocessors, vector processing, and a whole host of other technologies that show up in mainstream CPUs. Each chapter goes in-depth on the topic and shows you the kinds of decisions and tradeoffs that go into the design and implementation of a particular technology. At the end of each chapter is a section called “Fallacies and Pitfalls,” which shows the reader what to look out for when designing and implementing the technology just discussed. The chapters also have a “Historical Perspectives” section that gives the history of the development of the technology under consideration. Finally, an appendix contains a survey of various RISC architectures: MIPS, SPARC, PowerPC, and PA-RISC; there’s also an entire appendix devoted to the 80×86.

A book like this doesn’t come cheap though. Engineering majors are used to paying $80 a piece for textbooks, but others might balk at the price. Still, if you want the best then you’ve got to pay for it. I refer to this book constantly while writing my articles; if you check the bibliography section of most of them you’ll most likely see Hennessy and Patterson listed somewhere. So if you’re serious about learning about computer architecture, and you’re willing to invest the time, effort, and money, then this book is essential reading.