TechBulletin

The Pentium Chronicles - The People, Passion And Politics Behind Intel’S Landmark Chips

Robert Colwell, the project manager of Intel, jots down his experiences and experimentations in guiding the design team of the famous microprocessor Intel-P6. The author goes down his memory lane, presenting his creation in the domains of engineering and managing without transgressing the layman’s realms. Such a perfect miscellany of his experiences finely blended with humor and reality is a sumptuous feast for engineers and project managers.

Behind every successful project, there’s a team that racked their brain for the conception and behind it laid incidents and experiences that are sweetly recalled by the members.

 
The brainchild of Intel the P6 has a story too, narrated by its project manager Robert Colwell. The author joined Intel’s new Oregon microprocessor design division in the summer of 1990 as a senior computer architect of the P6 project. After two years he was to get promoted as the architecture manager and to serve as the lead IA-32 architect from 1992 to 2000. Inclined towards a target to beat the competitor’s design and overwhelm the predecessor design, the whole team grew out with Colwell as the genesis. He happily narrates his first days in the corporation with a subtle sense of humor, especially when he asks his boss “What’s P5? Can you tell me more about Intel’s process technology plans? And where’s the bathroom?” P6 was the magnificent work of art of the Intel Corporation which became the Pentium Pro microprocessor for Pentium II and III and the prototype for the Xeon and Celeron variants. Extrapolating the golden rule of Moore’s law that “the number of transistors per square inch on integrated circuits had doubled every year since the integrated circuit was invented”, P6 was to have 8 to 10 million transistors. This improvement Colwell says “is a double edged sword” since the target is not stationary as any new chip has to beat the competition’s new design and the process-migration chips from any source, which may include the mother company itself.

The team’s design modus operandi was “proliferation thinking” by which a new micro architecture was conceived and later a chip was designed and produced in limited volumes. In this proto design, every transistor is spent effectively and additional transistors are utilized for additional performance, making the first initiation plausible and thereby making money out of the successive proliferations of the original chip by milking the benefits of Moore’s law. Reusing to the maximum extent the available existing solutions, tools and methods, addressing in parallel the special technology challenges in circuits, buses, validation and multiprocessing was the approach undertook in the design process which proved to be fruitful. With a team comprising members from diverse backgrounds ( most of them were recent college graduates), the majority had no idea about Intel x86 chips or no work experience at Intel which the author says had been helpful in providing the right balance of new challenges and answers. Moreover he emphasizes enormously on the wealth of “senior wisdom” in his team, which was a group of experienced senior people adept at making rightful choices that guide the efforts of hundreds of others over the project’s life.  “Guard them!” he says as they are irreplaceable and their intuitions are essential to steering the project onto the right track. Unlike small projects, large projects should not be given the ad hoc treatment which would ruin the efficiency. While structuring and scheduling the project, Colwell came out with a framework consisting of four phases – Concept, Refinement, Realization, and Production.

In the first phase, the request is satisfied by a plethora of ways and brainstorming, in the words of the author always does well. Weeding out the ambiguities and prioritizing the solutions is carried out in the refinement phase in which the ideas are concentrated more. The actual engineering process is the Realization in which the team segregates the best idea budding out of the refinement phase and the prototype is constructed. Production forms the last phase of the process where the lay out developed hitherto is created into bulk volumes and this phase has little relevance to the design teams. All these four phases overlap in any project framework and this model is an extremely useful management tool adept at coordinating the design team. 

Birth of a concept is a very giant leap ahead in any project as it forms the genesis of the project. Colwell here welcomes inspiration, opportunities and of course errors. And remember, successful project emerges out only if you get the basics right .Goal setting is a very basic yet important phenomenon as it forms the foundation of the project. Assuming that all the team members have the same implicit goals should be avoided and the diversity must be accepted. Choosing the right man for the right work is yet another basic factor.  Clear definition of leadership and the delegation of authority must be set conspicuously.  Another funny concept that the author shares is “Kooshing”. Colwell always emphasizes the need for doers not talkers on a concept phase team. To warn and coerce monopolizing talkers, a small rubber ball called Koosh ball is tossed up and down during the meeting. This was to indicate to the loquacious talker to save his skin and sit down quietly. The best judgement of their team according to the author was their determination to focus on 32-bit performance which became a feather in the hat for the corporation. The author fondly recollects his customer visits to Microsoft in which he jots down his experiences with the Windows 95 and NT development teams, Novell and Compaq.

To err is human. It is also humanly possible to efface these errors as far as the domain of coding is considered. Colwell speaks out the necessity to build a design devoid of bugs and admits that he was frustrated with the bugs that had covertly crept in to the design.

What’s so amazing is that the author strikes a wonderful relationship between geology and the computer geology in which he specifies the Intel’s x86 architectural origins in the Paleozoic era of computer geology?

Colwell also emphasizes more on the people factor in the team that had a size of nearly 400 design engineers. His sense of humor is conspicuously shown in his experiences in recruiting and firing employees. He answers the questions that were thought-provoking, triggered by people who met him in diverse locations like “Was the P6 project affected by the Pentium’s floating point divider bug”, “why did you leave Intel” and so on. Also he shares some of the Pentium jokes, going in to the shoes of a standup comedian. Finally he justifies his leaving Intel and his views on design engineering which shows his expertise and experience in that field.

We would like to express our heart-felt thanks Stacy Smith from John Wiley & Sons, Inc. for giving us the privilege to review this marvelous piece of computer history.