Designing the 8080, By Masatoshi Shima

"With the 8080, Intel wanted to develop a second generation 8-bit microprocessor which would compete with 16-bit minicomputers", explains Shima, "I knew how to do it."

Shima had gained experience of minicomputers at Ricoh to add to the experience of business systems he had gained at Busicom.

With Intel's engineers heavily weighted towards process engineering, logic experience was scarce, and systems knowledge was scarcer still.

"I joined Intel in November 1972. By the end of the year I had written the functional specification of the 8080," says Shima.

The logic design was done within three months - all by Shima - and he also did the circuit design. "I then took the project to development with two mask designers," says Shima.

"The architecture of the 8008 had been good, but its performance was not good. They didn't take a systems approach with the 8008", he says, "that's why Intel was not successful with the 8008 - they copied the 4004's idea without recognising the system requirements."

As a result of that, the 8008 used a large number of support chips - about 20.

Shima set out to design the 8080 "with almost nothing needed in support". Six chips were required for the system interface of which Shima developed five (8255 I/O, 8251 USART, 8253 Timer, 8259 Interrupt, and 8257 DMA) - most of which were used in the IBM PC when it came out in 1981.

The 8080 project was completed on August 9th 1973. On August 10th, Shima's wife gave birth to twin daughters. In February 1974, Shima presented the 8080 at the International Solid State Circuits Conference.

When it was put on the market, in April 1974, there were so many orders for the 8080 that Intel recovered its development costs from the first five months of shipments.

Shima noticed another symptom of the 8080's success: "After the 8080 became successful many big US companies visited Intel, instead of Intel visiting big companies!"

The widespread and varied interest in the chip also changed Intel's perception of the microprocessor business, reckons Shima:

"Before the 8080 the judgement of Intel management was that, even if Intel took ten per cent of the market, that would be less than 1,000 units a year - because they were positioning the 8080 to be a minicomputer - many other engineers thought the same. But not me."

NEXT FRIDAY, August 22nd: Founding Zilog, by Federico Faggin