I did Assembler programming on this chip - I still have the reference book. It was definitely ahead of its time.
There were some comparable chips in its era, like the NS32016, but it was ahead of the competing Intel 8086 easily.
It wasn't until the 386, 7 years later, that Intel got a comparable instruction set, however Motorola's was even then much nice to work with.
Unfortunately the lovely instructions set turned out to be an anchor. x86 is butt ugly but by a stroke of luck all instructions were of the "Read-Modify-Write" type, allowing Intel to implement a CPU with a RISC like core. I.e. Intel's complex instructions could be split up into simpler instruction that only did one thing, so instead of one instruction doing "Read, Modify and Write", you got three separate instructions doing only one thing and one thing well.
Motorola couldn't do this for the 68k. To remain competitive Motorola opted to go for a new architecture (
PowerPC). As a last hurrah Motorola reworked the 68K into the ColdFire CPU. The
ColdFire could be clocked much higher, but at the cost of software compatibility. Atari and Amiga couldn't use the Coldfire and died off, while Apple went with PPC for their macs.
Why did the SNES used an 8-bit processor instead of this?
Surely it would have been cheap to manufacture by the last decade of the century.
Two reasons. It's easier to port code from the NES to the SNES CPU, as they are much more similar, and they saved a few cents by having a simpler main board. A 8-bit bus requires less traces and pins, which is ultimately cheaper.