The microprocessor is 50 years old: it all started with the Intel 4004

2300 transistors, 108 kilohertz clock frequency and 4-bit arithmetic unit: the modestly equipped Intel 4004 is the world’s first microprocessor on a single piece of silicon. Intel had developed it for the Japanese company Busicom, which used the 4004 in the 141-PF desktop computer from 1972.

In 1971, it was not foreseeable that the successors to the 4004 desktop processor – that is, the 8008, 8080 and 8086, but also, for example, the Zilog Z80 – would become the most important engines of the numerical revolution. Even Silicon Valley, where Intel, AMD, Fairchild and HP were already located at the time, did not have its later importance by far.

The 4004 belonged to a chipset called MCS-4, which also included the 2048-bit ROM component 4001, RAM chip 4002, and I/O component 4003. Intel’s main developers were Federico Faggin and Ted Hoff, Stan Mazor and Masatoshi Shima, who moved from Busicom to Intel in 1969.

At the time, not only Intel was working on the “microprocessor on a single chip” concept, but also Texas Instruments (TI) and some arms companies. However, the Intel 4004 was ready in the summer of 1971, was – as mentioned at the beginning – announced in November 1971 and was also used in the Busicom 141-PF.

Intel announced the 4004 in an ad in Electronic News magazine.

(Picture: Intel)

Microprocessors existed before the Intel 4004 and TI TMX 1795, but not combined on a single chip. But there were already logic ICs such as the famous 74xx series, which TI had been producing since the 1960s. Such “transistor-transistor logic” (TTL) chips, which provide simple digital basic functions such as NAND gates , can be interconnected to form arithmetic units. But it took a lot, and computer boards with dozens of TTL chips were also called TTL tombs.

In comparison, single-chip microprocessors were not only more compact, but they also consumed much less power. Due to the rapid development of chips and manufacturing technology, microprocessors quickly became faster or, with a similar range of functions, cheaper and more economical. The significantly more powerful Intel 8008 8-bit microprocessor with 3500 transistors and 800 kHz appeared already in early 1972.

Microprocessor development from the very beginning was anything but simple. Because Intel 4004 and 8008 could not be used flexibly enough for a long time. Federico Faggin and Masatoshi Shima therefore developed the 8080 with 6000 full-speed transistors until 1974.

Intel 4004 in DIL package with 16 pins.

(Picture: Intel)

Following the oil crisis, there was a recession in 1974, to which Intel also responded with restructuring. They didn’t match Faggin. He left the company and founded the Zilog company in 1974 with Ralph Ungermann in Cupertino, not far from Intel’s headquarters in Santa Clara. Masatoshi Shima followed later. In 1976, Zilog released the Z80, which was widely used from the Sinclair ZX80 to graphing pocket calculators such as the TI-81.

Motorola introduced the 6800 8-bit processor in 1974, based on the PDP11 architecture from Digital Equipment Corporation (DEC). The line continued in 1976 with the 6809 and 6801. The Motorola 6801 with 35,000 transistors and the 6809 with 40,000 transistors played in the higher leagues. The 6809 had a hardware multiplier. The Instruction Set Architecture (ISA) was 16-bit.

In the meantime, however, another “6800 descendant” had emerged, the similarly designed 6502 from MOS Technology. This is where Motorola developer Chuck Peddle had migrated. The 6502 was extremely popular, first in the Atari 800, Apple I, II, Commodore PET and VC-20, then later in 1982 as the 6510 in the C64 “breadbox”.

However, in 1979, the 6800’s successor, the Motorola 68000, became particularly important. Its computer hardware (ALU) was only 16-bit, but the ISA and all registers already had forward-thinking 32-bit. The linear address space was then a gigantic 4 GB, of which a whopping 16 MB was initially still physically usable. Chief architect Skip Stritter had previously worked on the IBM 370 mainframe microprocessor, so a number of ideas from the /370 architecture were incorporated into the 68000.

In 1977, Intel improved the 8080 with the 8085 in response to the Z80. But at that time, a 16-bit processor was already late, because the competition had already finished it or was in the works. Intel therefore needed a cheap 8085 as soon as possible, but it had been drilled to 16 bits – and the 8086 processor became that last resort.

1978 Intel 8086: 29,000 transistors with 3-micrometer structures, 5 MHz clock rate

(Picture: Intel)

The 8086, introduced on June 4, 1978, inherited much of the 8085 and continued to fit in an inexpensive 40-pin DIL package. Getting by with a few external connections, however, required some tricks and contortions, some of which even years later drove programmers to despair. In order to save pins, the address and data signals were not only switched one after the other on the same connections by multiplexing.

The 8086 was more of a makeshift 16-bit build, but in the ancestor of all modern x86 PCs, the IBM PC, an even smaller 8086 version was used in 1981: the 8088 developed by Intel Israel in Haifa. Its most important “advantage” was the reduced external data bus from 16 to 8 bits. This reduced performance, but the computer hardware required in addition to the microprocessor could be greatly simplified and configured at lower cost. It is well known that the IBM PC should not be particularly fast, but should at least be particularly cheap by IBM standards. That’s changed a lot in the 40 years since then – but that’s a new story, that of the x86 PC.

Learn more about the history of microprocessors:

More from c't magazine

More from c't magazine

More from c't magazine

(waiting line)

Comments are closed.