A test bench in a new computer firm in Cupertino in California on a spring morning, a large green circuit card forty centimetres wide, along its middle three empty chip sockets, a hand in a rolled-up pale blue shirt sleeve pressing a long white ceramic chip five centimetres long with a square gold lid and two rows of short gold legs into the first socket, two more identical chips on a square of grey foam, every chip plain smooth and unmarked, a grey cable running to a cream-coloured terminal with its dark screen turned aside, through a wide window rows of apricot trees in white blossom, morning light from the left.
What it was used for
The AL1 was an 8-bit MOS processor chip from Four-Phase Systems, never sold on its own but built into the company's computers. By the account at cpushack it held about 4,000 transistors, or over 1,000 gates, in 10-micrometre technology with a four-phase clock, had 8 registers including the program counter and a full 8-bit ALU, ran at about 1 megahertz and sat in a 40-pin package that spared it from multiplexing its connections. In the System IV/70, 3 AL1s together formed a 24-bit computer, supported by 3 ROM devices, 3 random-logic devices and the clock generation.
What was new
Whether the AL1 was the first microprocessor remains unresolved, and the dispute over it has entered the record. The Computer History Museum catalogues under number 102770290 a 7-page document prepared on 3 April 1995 by Lee Boysel, founder of Four-Phase Systems and designer of the AL1; it was meant to show that the AL1 incorporated claims that Texas Instruments filed for patent only several years later. The barb: the device was made in 1969 so that Four-Phase could sell computers, not chips - it did the firm no good in the component market. It became known only decades later, as evidence in patent disputes.
What the sources leave open: The technical data come from an enthusiast site, not from Four-Phase or a museum; the same page contradicts itself on die area (130 by 120 mils against "100mm2"), so that figure is omitted. The Computer History Museum record names no case number and no party; cpushack mentions only a 1990 court case in which Texas Instruments cited the AL1 against an Intel patent. No quotation from Boysel was found in the original, so none appears in the text.