#PDF scan of the #paper on the design of the 20-bit bit-serial “MP944” CPU from 01971, part of the #CADC, which sure doesn’t sound like a microprocessor to me, despite the domain name.
The multiplier (“PMU”) was a 28-pin DIP, the divider (“PDU”) was another 28-pin DIP, the “special logic [functional] unit” was another 28-pin DIP, and the “steering unit” (of which there were three, one for each of the previous three modules) was another 28-pin DIP, and there were 19 vertical-microcode ROM chips: “Each functional unit is controlled by its own micro-instruction ROM.” To me it sounds like there were at most three CPUs, each consisting of a steering unit and either a multiplier, a divider, or a “special logic unit”, and maybe just one CPU split across 25 chips; when the paper uses the term “CPU”, it seems to be referring to all of these functional units together and perhaps the ROMs as well, not just one of them.
Moreover, the multiplier and divider just blindly did a multiply and, respectively, a divide, every “frame”. Only the “special logic function” could do “Conditional, and Unconditional data branching”. And all three “functional units” were controlled by some control bits from the “system executive control”, a group of ROM chips.
The “steering unit” was a “three channel serial digital data multiplexer” with three outputs which could add or subtract data under the control of part of the instruction word on its way to the PMU, PDU, or SLF.
It seems like the program counter, or rather counters, were located in the microcode ROM chips, which could increment them or reset or reload them under the control of some control lines. #hardware #architecture #history
on 02026-05-26Purportedly the first microprocessor was Ray Holt’s 20-bit bit-serial MP944 from 01971, part of the #CADC, developed for the F-14 naval fighter jet, but the Navy classified it until 01998. #history
on 02026-05-26