Note – I’ll be busy working on some personal projects for the next few days but when I return I share something that you may find even more interesting than the technology IBM used to develop The Cell – an old and unanticipated way of working that gave the designers an unparalleled advantage in drawing complex micro circuits that have been the “brains” of computers they have sold and deployed all over the globe.
Welcome to my autobiography.
I am back after one month and in this post will share how designers in IBM’s Systems and Technology group working in Rochester, Minnesota worked in a way that seemed contrary to anything I had ever seen.
The Graphics Lab
After nearly two months of background checks, travel and housing arrangements and reviewing the fundamentals of CMOS chip design, I approached IBM’s sprawling Rochester, Minnesota technology campus on a spring morning in March of 2004. After checking in with security I was directed to the graphics lab in building 7. Passing through two more secured doors I entered a 24 x 36 square foot room known as the “graphics lab”.
“Welcome!” the lead mask designer boomed as I entered the room that seated no less than 17 mask design specialists. The least experienced mask designer I met had been with IBM for 7 years. Two had been at the same desk – a design workstation console – for 17 and 23 years, respectively.
The Physical Environment of The Graphics Lab
The room itself looked like something out of the eighties, with raised, white floors, odd-shaped desks and IBM mission-statement posters featuring giant mainframe computers from the bygone era.
The room was absolutely FILTHY, in spite of daily service by a janitor who was very reliable but not interested in anything except a very superficial cleaning – since either mode paid the same. Stains and debris from over 30 years of spilled coffee, sausage and eggs breakfast eaten at desks and pot-luck chili and venison brought in to feed folks during occasional late-night working sessions.
In this room, I estimated, was seated the accumulated chip design experience of over 140 years; most of it on the same spot on the floor – even if in a chair of a more modern design, from every few years! At each workstation location in the lab, tracks on the floor were clearly visible from the casters on their chairs; tracks that the janitor’s mop had wiped-over, but not completely removed. On the curved desks, Gnomes, miniature muscle-cars, refrigerator magnets, trophies of walleyes, engraved-brass employee service award plaques and Heinz ketchup squeeze packets could be found. Once an object was placed on the desk, it seemed, unless it obstructed a path from the fingers to the keyboard, it remained forever, becoming part of the magnificent but oddly eccentric legacy of the graphics lab and its IBM mask designers.
During the 1980’s when semiconductor design technology facilities were cropping up around the country it was common to see mask designers working in one room – sometimes with the lights dimmed to enable the designers to see their display screens better. But as the use of cubicles (which were often integrated with tall privacy walls) became more wide-spread, many darkened mask design lab rooms went away, and with it, in my view, a more informal, conversational, collaborative mask design working culture also disappeared.
The Mask Designers and Their Work Style
This culture still thrives in the graphics lab at IBM, in Rochester, Minnesota. In fact, I have come to believe one outcome of this culture is a kind of “mastermind” that is formed when intelligent, highly creative people work together in close proximity (but it can also be a virtual work group). Unseen yet powerful, this mastermind then acts to help guide and direct the activities of the mask designers who work to create complex integrated circuits and improve methods to produce and check the correctness of them more effectively than one person could do so on his/her own.
I personally witnessed this phenomenon many times in the last quarter of 2004, when working in this lab with these clever, highly experienced and eccentric mask designers in Rochester, Minnesota. Information about the interpretation of a design rule specification or the verification of a circuit drawing were often shouted-out, then provided in real-time. Questions of problems that could only be posed by first looking at the graphics screen that was displaying the information, were quickly addressed by the questioner walking a few steps and getting help. What one learned, all learned, simultaneously. Every interested or vested participant could be included on breakfast and exercise walks, or to and from training or communications meetings, whether planned or spontaneous.
Even though the graphics lab at IBM, in Rochester, Minnesota seemed like something out of the “dark ages” with its 1980’s decor, grimy appearance and mix of eccentric personalities that sometimes clashed, its collaborative work culture offered many tactical advantages over more contemporary mask design office set-ups.
Nearly every advantage of effectiveness or efficiency offered in this kind of office was offset by a corresponding disadvantage – at least for some mask designers. For example, since everyone working in the large room was in sight of everyone else, there was no privacy. Since the expectation was that everyone would walk, talk, eat and share resources together, any person who did not go along, was perceived as “different”, no matter how unfair. Lastly, social and political hierarchy was present and active whenever members of the group participated in work activities. I saw this almost every time the group attended lunch together. The senior mask designers alway led a single file line through IBM’s endless hallways to the cafeteria. When the senior members of the work group completed eating their meal and rose to exit the cafeteria after their lunch, nearly every time, all members took their cue and also left the table.
In spite of occasional uncomfortable feelings associated with the open group office setting, on balance, the advantages outweighed the disadvantages, in my opinion.
But did the informal, collaborative work-style in Rochester’s graphics lab help IBM workers work smarter and produce more complex, higher performance chip-set architectures than those from other groups or at other companies?
In the end, I learned to use the peculiar chip design software the good folks inside IBM’s “graphics lab” used. Already 20 years into my mask design career, I was a complete novice (and the least senior in the lab). But with the help of others, I was able to complete a small assignment and finish the contract strong in July of 2004.
I am grateful to have been welcomed into this somewhat odd assortment of individuals, somehow stuck in time but yet years ahead of the competition in the race to work smarter.
In the next post you will read about a phone call from David – a mask designer friend – and another invitation, which turned-out to be quite special because of where the call originated.