A witty and highly informative talk, about his first (the first?) Gallium Arsenide based computer, the Cray 3. As of the lecture, late 1988, he hoped to deliver the computer before the end of 1989 (but that didn’t turn out.) Lots of detail on packaging technology and the progress of integration. The lecture also covers Cray’s personal career moves and the series of ever-faster machines he architected, and has a bit of a look forwards to the Cray 4, as then planned. His machines kept getting physically smaller, and were dominated by cooling.
He quips that it’s very hard to make curves go upwards on a log scale: