Customer Reviews

The Pentium Chronicles: The People, Passion, and Politics Behind Intel's Landmark Chips (Practitioners)

5 star:		(5)
4 star:		(4)
3 star:		(1)
2 star:		(1)
1 star:		(1)

 

 

I rated this book as three stars because the title promised
more than it delivered. Nevertheless, it is a very good book,
especially for people who have to manage huge projects in
complex technical areas. Colwell clearly is skilled in
technology, and has tremendous insight and experience to convey.

My expectations were different. Tracy Kidder's Soul of a New
Machine created excitement and tension into the development of
computers, at least as of the early 1980s. The machine was
successful for Data General to some extent, but faded as a blip
in history that few remember. Intel's P6, Colwell's baby, is
totally the opposite, selling hundreds of millions of copies in
multiple forms since its inception. Kidder spins magic about
the development process. Colwell tells how to make it happen
-- no magic, just cleverness and grunt work.

What I found most valuable were Colwell's methods for taking
on this huge project. Quantify your goals, quantify the merits
of each idea, and quantify your progress toward the goal.
Without these measurements, you have no idea when you will finish
and whether you will succeed when you get there. In
a field where technology moves very quickly, the difference
between success and failure is not so much if you complete
the job, but when you complete the job.

Colwell pulls some punches because of corporate and personal
sensitivities. He does not tell us very much about the
P6 processor, but what is revealed is done skillfully in
layman terms so that the nontechnical reader can follow the development.
For something this complex there has to be dozens of interesting
facets whose challenges and successful development would
make for a good read. Out-of-order execution, that is, the
ability to reorder program instructions on the fly in order
to boost performance, gets the biggest play. Chefs prepare
courses out of order when they prepare elegant meals. Desserts
may be started first and eaten last, if the dessert preparation
time exceeds the entree preparation time. Colwell uses his own
analogies to explain this, and it works quite well.

Apart from out-of-order execution, very little else is developed.
It may be that Kidder's book actually has more technical content,
but I have not taken the time to make the comparison. The
point is that Colwell comes up short in the technical arena,
and this is a disappointment.

The fact that Colwell is both the author and the project manager
makes for a sharp contrast with Kidder's work because Kidder
was an outsider and was not responsible for technical decisions,
whereas Colwell expends many pages defending his design
choices. The personal involvement detracts from this defense,
because it leaves the reader wondering if the defense is
self-serving or is actually a fair appraisal of the facts.

Sensitivies come to the fore in the final pages when Colwell
describes recognition as well as litigation arising from
the P6 project. Perhaps thousands of people participated
in some fashion over a five year period, and only a few were
heralded by Intel. How do you select them, and how do
you placate the wannabes?

The litigation relates to patent
infringment allegations, which Colwell equates to accusations
of theft of ideas. In reality, patent infringement does not
necessarily involve theft of ideas. The patent system is such
that people can develop products whose infringement with
patents cannot be discovered until after the
products are on the market. (This was the
case in the 1990s when patent applications were secret
for the several years between their date of filing
and the date of their issuing. Currently, patents are
made public immediately after filing, even though
they are not valid yet. This helps reduce accidental
infringement.) Colwell believes the infringement
suit against the P6 that he cites soils his reputation for
honesty. It really does not. Intel licensed the patent
in question, with words that credited the inventor for
contributing to the P6. This incenses Colwell because
the P6 design was done independently of the licensed patent.
An outsider would probably relate the facts with words
less charged with emotion, and the book would improve.

To sum up my views of the book, it has major strengths in
management techniques, and falls short in entertainment value.
The Kidder book as well as Michael Malone's book entitled
The Microprocessor, A Biography, are two that are high in
entertainment that I would recommend ahead of Colwell's book
for reading on the airplane. On the other hand, if your
company has just selected you to move forward its technology
with a horizon of five to ten years, you had better put
Colwell on your required reading list.

If you are involved in the world of integrated circuits, or considering becoming involved, then you'd be crazy to pass up this book. It's no less than a first-hand account of how the golden age of Intel came to be, as well as how it came to a close.

In the early 90's, the common wisdom in the CPU industry was that a buzzword-complete (out-of-order, superscalar, superpipelined, speculative execution) x86 was simply impossible to sucessfully execute, hence the smorgasboard of then-new competing RISC architectures. The book's author led the architecture development of the project that proved otherwise.

What's truly astonishing about a project of this scale is the vast array of things that have to go right in order to prevent a catastrophe (or, as a colleague says, it's not the rocket science, it's the rock science). Even more amazing is how many things the P6 team fundamentally got right (at least according to my own 15 years of IC experience). I was also delighted to find simple and yet brilliant ideas that were new to me, such as assigning cubicles by overlaying the building floorplan with the chip floorplan.

The parts of the book that I found most entertaining (from the outside looking in, that is) were descriptions of the naive attempts to replicate and exceed the success of the P6 project, largely by deprecating the very mechanisms that led to that success.

A word of warning: If you don't already have a lot of experience with large projects, you'll probably have to resist the urge to disbelieve many of the anecdotes. Obstructing Pentium 4 engineers from knowing their own plan seems ridiculous, but I can assure you that in my years in the IC business, I've seen worse.

Because of the exponential upward trend in realizable complexity, as well as the cost of tapeouts, it seems the future (if not the present) of the IC industry to be predominated by large projects such as the P6. If this is the career path that you have chosen for yourself, then you would do well to heed the lessons taught by this book.

As may be likely with many readers, I first heard of Mr. Colwell from his IEEE Computer columns, of which I was big fan. Several times I was tempted to send him comments about the insights and occasional humor, but I never did. The same kind of writing and attitude is clearly on display in "Chronicles".

Because other reviewers mentioned "Soul of a New Machine," I agree that it is hard to escape, even though the two books are much different. When I was a young engineer in the early 80s, "Soul" was (and is) a great book. I did not have the experience to grasp some lessons to be learned from the story, and a 1980s version of "Chronicles" would not have hit home, either.

However, Mr. Colwell is completely believable in his anecdotes and in the presentation of the big picture, the project, and countless details both technical and personal. Experienced engineers will no doubt see themselves, their colleagues, and their projects in one form or another. I don't mean just the "Dilbert" moments and inevitable personal clashes, but also the serious business and technical challenges that any complex project must face. I found myself nodding in agreement again and again with his conclusions and advice.

Two reasons I liked Mr. Colwell's columns are that he is not just a techno-geek and that an interesting feisty personality showed through. You see that feistiness at times in the book with stories of taking on the powers that be, for better or worse, without feeling like you are listening to someone covering his tracks to make himself look good. The true engineer comes through, with enough polish to be around executives and to be allowed with customers.

It's really a treat. It should be in the bookshelf of a grad student or practioner in the fields of computer and electronics engineering.

It is not written by some professional writer, instead, it is a book written by a engineer, with the very first-hand experience of the development of P6 architecture. If this book should have some imperfection or even mistake, I will take a psychological perspective to anylyze it -- just like I analyze my own development memos. 8-) As a reader of At Random, I can say that the writing of Mr. Colwell is comparable with those professional writers.

BTW, I think the subtitle is quite fit for the content of this book.

I thought of this book as a tome of helpful advice for any project manage that has to work on a project that veers far into the unknown. The author was the Chief Architect for the P6 (Pentium Pro) microprocessor. This was a radically new type of processor that broke from previous chip architectures. The team was required to explore untested ideas but at the same time bring them to an operable design. This required considerable balancing and deft handling of various stumbling blocks. The author presents a lot of observations on how to accomplish this. It was also very interesting, for a person who is interested in microprocessors, to see where exactly time is spent in these massive projects. The author gives lots of technical details, but I would have like a bit more. Of course, the intention of this book was as a project management guide, so that isn't necessarily bad.

This is an excellent case study about Intel's P6 project -- that became the Pentium Pro and part of the lineage for a number of other processors. The author provides the story of compromises made while the project proceded and the corporate politics endured at Intel. He also discusses the Pentium flaw episode and provides a technical rationale that thay may have led to that unrelated debacle. It is fun to read about some of the technical aspects of this project. One flaw to the story is the lack of additional information on why Intel settled out of court and paid millions of dollars for alleged patent infringement accusations relating to the P6 project. The book is surprisingly upbeat considering the author no longer works at Intel. If you like this book, Brian Bagnall's "On the Edge" about the rise and fall of Commodore computer is an excellent follow-up.

I'll give this book five stars, because it captured me. Bob is great at writing: I read him as a columnist for 4 years and I really enjoyed this book. I particulary liked the tactics and advices on how to make it happen, on how to build a large team that really works. The only thing that disappointed me is the lack of very inside informations about the chip, the design solutions, the "eureka" moments about out-of-order execution: the creative moments, in short. The management part is great, but I miss a part more focused on the engineering side (however the title is clear: people, passion and politics)

It is rare for a book to be released chronicling the development of a chip or computer. The best known example is Tracy Kidder's "Soul of a New Machine", some 20 years ago. Colwell might be considered to offer a sequel. He explains how his group at Intel designed and produced several generations of Pentiums.

The narrative is drier than Kidder's. But goes into far more technical detail. Frankly, if you are not already an electrical engineer, then it will be hard to appreciate many passages.

People in the semiconductor industry should scrutinise this book. To you, the hardware problems will be familiar. The attraction is in how a large group was mobilised and managed, in delivering a very complex chip on time. There may be useful management tips for you.

On the first page of this book Dr. Colwell gets his marching orders from his boss: 'Your job is to beat the P5 chip by a factor of two on the same process technology. Any Questions.'

'Three,' he replied. 'What's a P5?' What's the process technology plans? Where's the bathroom.'

The P5 became the Pentium chip. The process technology doesn't matter in this book as it was the same. And he found the bathroom.

Dr. Colwell's new chip, the P6 came to the market first as the Pentium Pro, the same basic design was subsequently modified as the basic core for the Pentium II, Pentium III, Celeron, Xeon, and the current Centrino products.

But the details of the chip aren't the strong point of the book. It's the organization and structure of a how a big development team works. From the little details like finding an unused storage room to use as a conference room where the blackboards wouldn't get erased, to hiring, firing and getting the product out.

I can't help but compare this with Tracy Kidders book 'The Soul of a New Machine' written twenty or so years ago. Kidder was a better writer, but Colwell was the one in charge, making the whole thing happen. I would have liked to see a little more technical detail, but I've been in this business a long time and have more interests along these lines than the average reader. As it is, it is an excellent book on project management and I enjoyed it very much.

If you enjoyed "The Soul of a New Machine" by Tracy Kidder, you don't want to miss this one. It will establish a new benchmark, I think, for much quoted books in the decade folowing its publication.