Design Cutoff

1978 · Multi-Project VLSI Chip · Work 906 of 1,871

A university design laboratory in Cambridge in Massachusetts on a December evening, black night in the tall windows, fluorescent light, students in sweaters bent over terminals along the far wall, their screens glowing plain pale green and soft in focus, on a table a flat beige pen plotter with four coloured pens drawing on a large sheet a dense pattern of overlapping rectangles and stripes in green, red, blue and yellow, paper coffee cups, in front a tall woman with long hair in a dark blouse holding up a large plot toward the ceiling lights, the coloured pattern glowing through the thin paper, everyone seen from directly behind, faces turned fully away.

Who
Lynn Conway (1938–2024) was a computer scientist and chip pioneer whose methods changed how microchips are designed and taught. Born in Mount Vernon, New York, she studied at MIT and at Columbia University and joined IBM Research. There she invented a scheme that lets a computer issue several instructions at once, out of order. When she began her gender transition in 1968, IBM fired her. With a new identity she started over in early 1969 at the bottom as a contract programmer, became a computer architect at Memorex and moved to Xerox PARC in Palo Alto in 1973. With Carver Mead of Caltech she developed simple, scalable rules for designing very large circuits and was principal author of the textbook "Introduction to VLSI Systems". Her 1978 course at MIT, and the network system she then built at PARC, carried chip design to many universities and led to the MOSIS service. At DARPA she planned the Strategic Computing Initiative, which also aimed to advance weapons systems. Later she taught at the University of Michigan. From 1999 on she made her story public and became a voice for transgender people. She died on 9 June 2024.
The moment
In the autumn of 1978 Conway taught a new course at MIT as a visiting professor: 32 students were to design chips of their own, joined by nine faculty and staff auditors. In the design lab they entered their layouts at DEC-20 terminals and had them drawn in colour on HP pen plotters to find mistakes. According to the course schedule, Tuesday 5 December 1978 was the design cutoff: all project files had to be at Xerox PARC for merging into one shared chip the next day. That same day Carver Mead spoke in the seminar. The files went to Palo Alto over the ARPANET; Conway recalled sending the final ones on 6 December. At PARC, 19 projects were merged into a multi-project chip, and a Hewlett-Packard laboratory made the wafers. On 18 January 1979 the finished chips were shipped back to MIT.
The other side
For Conway the course was a risk. She lived, she later wrote, in dread of being outed about her past; until then she had been sheltered in the laboratory, now she had to face the public. She kept her work at IBM secret. Intel's leaders Noyce, Moore and Grove, she felt, probably took her for Mead's helper with the book rather than the primary architect of the new method; women, she wrote, were outsiders in the semiconductor industry. Only in 2009 did the IEEE Computer Society honour her early work at IBM as well, with its Computer Pioneer Award.

What the sources leave open: No source says whether Conway was in the design lab on the evening of 5 December; her leadership of the course and the sending of the files are documented. The course schedule gives 5 December as the cutoff, Conway's later memoir 6 December for sending the files. Conway is seen from behind because there is no public-domain photograph of her; how she looked in 1978 is not described.

Further reading

Sources