Lynn Conway (1938–2024)

Lynn Conway was an American computer scientist, electrical engineer, researcher and university professor whose work at Xerox PARC in the 1970s led to the integrated circuit design and manufacturing methodology known as Very Large Scale Integration, or VLSI, something which touches very many facets of the world that we live in today.

In 1979, with Carver Mead, she published Introduction to VLSI Systems, a textbook that stayed the definitive bible of integrated circuit design for more than a decade.

This publication praised her merits in 2001 when she was 63 years old and in 2004 when she was 66. On June 9, nearly 20 years later, at the age of 86, Lynn Conway has left this earth.

After having studied electrical engineering at MIT and Columbia University’s School of Engineering and Applied Science Conway joined IBM and eventually worked for its Advanced Computing Systems (ACS) project at Menlo Park, California. It was as a member of this project that she developed the key ideas behind what was then called dynamic instruction scheduling, one of the key components of out-of-order execution that is now pervasive in all modern microprocessors.

In California, Conway, still troubled by depression due to long time gender dysphoria, sought help for a gender transition. While her immediate superiors and colleagues at IBM were supportive of a plan in which she would resign and be rehired with her new identity, this was not something which IBM CEO Thomas Watson Jr. was comfortable with. So Conway was fired by IBM in 1968.

She completed gender transition and took a new name and identity in what she would later call her “stealth” mode. To Columbia’s credit, they gave her no trouble in changing the name on her transcript and diploma. With these credentials, but no previous work experience she could cite in a CV, Conway took various contract programming jobs, eventually landing a position as a digital system designer and architect at Memorex in 1969.

In 1973 she was hired by Xerox PARC to join the LSI Systems group (Large Scale Integration) under Bert Sutherland. At PARC, Conway conceived of the idea of a multi-project wafer in which a mask set could be shared across multiple designs. The idea of scalable design rules soon followed.

It was there that Bert's brother Ivan Sutherland introduced her to Carver Mead of Caltech, who was a consultant to PARC. It was the right place, the right time and a perfect combination of capabilities of two very different people. In 1979 in a speech given at the IEEE, Gordon Moore declared that "besides products containing memory devices, it isn't clearhow future VLSI can be used in electronic products". Processors where just to hard to do in VLSI.

Together they developed and formulated revolutionary and groundbreaking ideas, rules and methods for to do other things than memory, namely processors in VLSI. Both began to teach the first courses on the topic, Mead at Caltech and Conway at MIT, as a visiting professor from 1978–1979. These courses and their scripts in turn led to the aforementioned book and a row of connected projects to create a national and open infrastructure to put these ideas into action in which Lynn Conway was also heavily involved.

The most illustrating anecdote I ever found on the revolution Conway's and Mead's book set of, is in George Gilder's Microcosm. It describes a scene where Mead visits Zilog to consult with founder Frederico Faggin. Walking down the hall by chance they meet Masatoshi Shima, a leading figure of chip design since he had played a crucial role in the realisation of the Intel 8080. "Hey Carver," said Shima, "I want a copy of your book."

"Oh," Mead answered modestly, "it's really just for teaching students. It doesn't have anything to tell someone like you. It's not up to your standards." But Shima insisted, "I've just finished the Z8000. It's the last of the big mothers. No way we can ever design a chip like that again. Send me your book."

Other top logic and silicon engineers were less humble and continued for years with mighty chips with many bugs before they turned around. In the meantime Conway's work enabled unlikely "wonders", e.g. one where 2 employees in a rather small European company, Steve Furber and Sophie Wilson, not only designed but also successfully implemented the first version of the most successful microprocessor of them all.

All of this was necessary to master the design of 3rd gen microchips because their number of components grew exponentially (see Moore's Law). Without Introduction to VLSI Systems and all the good things that came from it, only a handful of exceptionally talented and educated people in the whole world where able to master the rising complexity even in the last generation of LSI. The development of microelectronics and computing would never have reached the speed and diversity it reached throughout the 80ies and 90ies of the last century.

In 1984 Lynn Conway was made chief scientist for strategic computing at the Defense Advanced Research Projects Agency (DARPA) and left PARC. One year later she joined the University of Michigan, where she served as professor of electrical engineering and computer science. In the late 90ies, when she was about 60, a computing historian researched early IBM patents in performance computing. Lynn Conway then came out of "stealth mode", was active in supporting transgender rights and seeking recognition for women’s technological achievements.

In 2023, 50 years after she was recruited by PARC, Lynn Conway was recognized by the National Inventors Hall of Fame for her work on VLSI. If you have made it this far, you even might like to read the story as told by Lynn Conway herself.

• VLSI Reminiscences
• The MPC Adventures