< <! Datapoint 2200 (architecture was the model for the 8008, was built
< < using LSI, not an 8008)
< This sounds bogus. the 8008 model was the 4004 (stretched data paths).
< The 8008 model was clearly unlike most minis of the time with the internal
< (to the CPU) return address stack or limited depth.
<
< Allison
My information concerning the Datapoint 2200 is from a posting in
CPSR-HISTORY (the link I have to the archive is now dead, anybody have a
good link to that archive?) by the Chief Tech Officer at Datapoint from 1969
to 1984, Vic Poor, who was directly involved in the Intel 8008 project. I
also have a copy of the excellent book "The Microprocessor: A Biography" by
Michael S. Malone, which describes the development of the 4004 and 8008
Intel.
There is lots of misinformation on the web, where you possibly got the
impression that that 8008 was a 4004 with a streched data path. The 8008
was a very different design, actually an implementation of a Datapoint
architecture designed by Vic and Harry Pyle of Datapoint (then CTC, or
Computer Terminal Corporation), which was a bit-serial computer design.
Both Intel and TI were contracted to design the chip, but Intel couldn't
deliver on time (also maybe too slow?) and TI's product was doa ("its noise
margin was so poor it could not be used commercially"), so Datapoint used
the design it had already implemented using MSI chips and put out their
"Datapoint 2200". Anybody got one of those? All the Intel micoprocessors
up through the Pentium III have their roots in this design. That is were
the real data path stretching has occurred! :-)
Interestingly, the notion that "Intel only got into the microprocessor
business to sell its memory chips" was at least in part based on the 8008
project. According the Vic Poor, they only took on the cpu chip project in
order to keep the memory business with Datapoint! Some in Intel thought
that the market for microprocessors was too small and that the real money
was in volume production of memory chips!
John
jlewczyk(a)his.com