GI CP1600; was: What are the really unusual or weird computers?

On 25 Jun 2007 at 17:52, Liam Proven wrote: Much of this is misleading. First off, only the first word of an instruction is 10 bits wide--the remaining 6 bits in my GI book is labeled "Reserved for future expansion"--that's the first time I've seen that actually spelled out. The second word of each instruction (usually an immediate value) is 16 bits; but there is an instruction (SDBD) that tells the processor to fetch two 8 bit values out of succeeding words instead. So you could get by with a 10 bit ROM, though it wasn't the most efficient. Since the processor identified M1 fetches, one could have used 16-bit ROM and dedicated the unused 6 bits for use by other devices, such as a coprocessor. There was a pin on the CP1600 that seemed to be made for this--PCIT--when asserted, it inhibited the P-counter from incrementing on an instruction fetch, so the CP1600 could be forced to "stall" until some external device had completed its function. There were some clever ideas in the architecture. 8 16-bit registers, with R7 being the P-counter and R6 being the (upward- growing) stack pointer. R6 post-incremented when referenced for a memory write and pre-decremented for a memory read. R4 and R5 were auto-increment registers. The opcode for JUMP and CALL was the same-- the instruction saved the current PC in a register before loading the new value. An unconditional jump was signified by specifying this register as R7--which caused the old PC to be overwritten without saving. The I/O was miserably slow, however. Pricing was very good for the time--the 1K price in 1975 was $40. The GIMINI development system could be had for $3500. At the time, GI billed the 1600 as "The World's Most Powerful Single- Chip Microprocessor". Cheers, Chuck