|
|
The 8008
Computer Terminal Corporation (later known as Datapoint) had visited Intel in late 1969 to
contract Intel to design and build a custom bipolar memory chip for their small
CPU, which was to be built using TTL logic components (as was customary in those
days). The architecture of this 8-bit microprocessor was done by CTC, and – like
all simple CPU’s - was very similar to the 4004 architecture. Hoff and Mazor,
based on the simplicity of the architecture, proposed that Intel build this CPU
in a single chip.
Intel entered into a
development contract with CTC in early 1970 and Hal Feeney was assigned to work
on the chip. Feeney, with the assistance of Hoff and Mazor, started the 8008
design several months before the 4004. Because of its head start, the 8008 could
have been the world’s first microprocessor. After a few months of slow progress,
however, the project was stopped. It was restarted after Faggin had finished the
4004 and was free to supervise the project. Feeney, now working under Faggin’s
guidance, did the detailed design using the same design methodology Faggin had
developed for the 4004, enabling its successful implementation.
|
|
The 8080
The 8080 microprocessor was conceived by Federico Faggin in the fall of 1971 and he completed the architecture, instruction set and basic features in early 1972. The project engineer was Masatoshi Shima who started the development toward the end 1972 and completed it after approximately 15 months, under Faggin’s supervision, and using the design methodology developed by Faggin, similar to the one that was used for the development of the 4004 and 8008. The contribution by Stan Mazor was limited to a couple of instructions to the instruction set.
A verbatim transcription of a portion of Federico Faggin’s performance review related to the 8080 microprocessor, signed by Leslie Vadasz and approved by Andy Grove, should help clarify the attributions. Here it is:
EXEMPT PERFORMANCE & SALARY REVIEW
Name: FEDERICO FAGGIN
Review Period: from 12/72 to 12/73
PRESENT SALARY: $ 2250/mo INCREASE: $250/mo REVISED SALARY $ 2500/mo
JOB DESCRIPTION
Department Manager, MOS Small Machines
SUMMARY OF PERFORMANCE REVIEW
Outstanding performance in both product definition and product development
PERFORMANCE REVIEW
“The most important accomplishment in microcomputers in the last year is the definition (done in ’72 by Federico) and development of the 8080. This part at the time of this writing is “allegedly” working. While we are not in production yet, the results are truly outstanding both from the technical and project managerial point of view. Federico’s ability to guide that development (project engineer: M. Shima) is particularly noteworthy; we are able to deliver samples in the time frame he predicted a year ago!”
The performance review makes it clear that the definition of the 8080 was done by Federico Faggin, in 1972. Following is some of the background story of the 8080 in Faggin’s words.
“I was hired at Intel in April, 1970 to design the four chips of the MCS-4,
whose architecture had been previously defined by Ted Hoff and Stan Mazor.
These chips all used 16-pin packages, and were destined for Busicom, a
Japanese calculator manufacturer. The MCS-4 included the 4004, a 4-bit microprocessor – the first in the world -- memory and I/O devices working as a
system.
Another extant project at that time was the 8008, an 8-b microprocessor
whose architecture had been defined by Computer Terminal Corp., and had
been adapted by Hoff and Mazor to fit into a 16-pin package.
One of my first reactions to the 4004 and the 8008 was that the choice of
16-pin package was seriously and unnecessarily penalizing the CPU
performance. However, at that time, the exclusive use of 16-pin packages
was a sort of religion at Intel, and since Intel was 6 months late in the MCS-
4 project, to change the architecture was unthinkable.
Intel had no design methodology for random logic design with the new silicon gate technology, therefore my first task was to develop the new
methodology in parallel with the design of the 4 chips – a very difficult task
indeed. Only after the methodology had been developed it became possible
to design the 8008, and I ended up leading the 8008 project as well.
In the late summer of 1971, I visited several potential customers in Europe
to promote the use of the 4004 and 8008 microprocessors, the first was
already in production and was soon to be announced to the general market, while the second was in an advanced stage of development. During this trip I received good customer feedback which did strengthen my ideas on how to architect a good microprocessor.
I started thinking about the architecture of the new microprocessor – the 8080 - soon after my return from Europe and completed it in early 1972, at
which time I wrote a memo to management describing my proposal and
urging the immediate start of the project. I conceived the 8080 with a
40-pin package and N-channel MOS (the 4004 and the 8008 used the slower
P-channel MOS) which, when combined, would reduce the instruction
cycle-time five times compared to the 8008, and would make the interface
with memory and I/O devices much simpler than the one required by the
8008 (the 8008 needed two dozen external chips to interface with memory
and I/O, increasing the cost and occupying precious printed circuit board
area). The 8080 turned the very limited 8008 chip into a good and fast
microprocessor, by adding many new instructions, new addressing
modes and better facilities, in particular a new interrupt structure, while
retaining machine-code compatibility with it.
One of the major hurdles I had to overcome was the use of a 40-pin
package which was resisted by management, despite its obvious benefits.
Since I believed that the competition would soon respond to Intel’s
offerings with better products, I kept on prodding and lobbying until I finally
won approval for the project.
The 8080 design eventually started 9 months later and the chip was
introduced in the market in March 1974. Motorola introduced their first 8-bit
microprocessor, the 40-pin 6800, six months after the introduction of the 8080, and I always regretted the loss of 9 months of market advantage, which offered a window of opportunity to the competition.
Fortunately the 8080 was good enough to withstand the competition, but
Intel’s lead was irreversibly shortened. This experience was an important
part of my decision in 1974 to start my own company, Zilog, Inc.,
the first company totally dedicated to the nascent microprocessor market.”
Copy of yearly performance review
|
|
|
