For reasons that are hard to pin down, there are remarkably few accessible books charting the development of the modern microprocessor. Every other area of computing has had light poked into its every recess, but the story of the most fundamental component of all has remained pretty much a mystery.
It could be the inherent complexity of the subject matter, or the inscrutability of the engineers that design them. Perhaps the impersonal nature of the companies that produce these miniature calculating furnaces simply defeat any attempt to personalise the tale or drag out a readable narrative. But there is a story to be told - look at an enlarged art lithograph of these extraordinary objects and its hard not to ponder their immensely expensive but obscure creation.
This particular book on the genesis of Intels groundbreaking P6 programme of the early 1990s, wont satisfy everyones curiosity but it does give us some worthwhile pointers. Written by the chief engineer who headed the project (which gave us the Pentium Pro, and derivatives the Pentium II, III, and 4), Robert Colwell, it is a book by an insider. An insider is always mindful of his friends and is always going to lens his account, however unconsciously, to suit sensitivities the reader is unaware of. But perhaps there is something to be said for insiders. They dont throw many punches but they were there in a way no outsider ever could be.
Colwell commits the understandable sin of describing the P6s invention as a technical and organisational exercise. If these chapters start to drag, they bring home one aspect of the P6 that was not obvious at the time of its appearance it was designed from scratch with reference to nothing that preceded it. This was not an evolutionary product, despite being in the same x86 family.
The P6 was a big achievement as it turned out and arrived in the nick of time. Software complexity had steadily increased, and by the mid-1990s the Internet suddenly arrived, heaping yet more demands on microprocessors. To think that only a few years before, the company had been producing a generation of chips the 486 being the best example - that looked steam-powered by standards of 1995.
There are a few morsels to savour. Colwell who has since left Intel is occasionally disdainful of company management who come across as distant and insecure. There is a sense of a bureaucracy bearing down on the engineers, who live at the sharp end of turning ideas into something saleable but get little thanks.
He briefly delves into the three big technical controversies of the P6 era the 1999 chip ID fracas (whereby each chip could be uniquely identified, arousing privacy concerns), the 1995 FDIV Pentium bug (an error in the chips floating point function), and the contention that the P6 architecture borrowed in important ways from the intellectual property of Intels then rival DEC.
The FDIV bug wasnt actually anything to do with the P6 the Pentium was a different design team - but Colwell happened to be the first person in Intels management to be made aware of it and that had political implications.
He recognises this an important moment for the company and, by extension, the whole chip industry.
An obscure design erratum like FDIV had become cocktail party knowledge. Microprocessors had entered the popular culture. I was sure this was an important change in the rules, but as yet had no idea what the new rules were.
The dispute with DEC cost Intel $600 million, though Colwell presents this as a price Intel paid to buy public peace rather than an admission of guilt. This caving into pressure clearly rankles with him as it reflected, however unfairly, on the integrity of his teams work.
In the end Colwell left Intel in 1998 out of what he describes in his understated way as frustration. At no point does he make working at the company sound like fun. This admission comes at the end of the book when it should really have come at the beginning. This is where he should have started his account. This is a worthy book, and worth a recommendation, but you cant help but feel that hes ended up writing a long introduction that skimps on the best bits. If Intel has gone wrong or is going wrong we want to know where.
The Pentium Chronicles: The People, the Passion, and Politics behind Intels Landmark Chips
Robert P. Colwell
Wiley Interscience, 2006