What are the really unusual or weird computers?

Christian Corti wrote: Not really -- the basic interpreter is written in microcode, but I wouldn't call that execution in hardware. Also, I'd argue that the microcode is really nearly an ISA -- there are no pipeline hazards, the uword is very narrow (20b in the 2200A/B/C/S/T, 23b in the VP/MVP). The only thing that makes it ucode like is that some instruction formats have control fields to manipulate independent behaviors, and that jumps are within ucode page. Long jumps take a bit of extra work. The power of the first generation 2200 (A/B/C/S/T) was roughly the same as a 4 MHz Z80, I'd estimate. The VP umachine was effectively about 5x-8x faster, and it was a lot nicer to program too. A factor of 2.7 was due to a faster cycle time (1.6 us vs 600 ns per uinst); a factor of two was due to the 1st gen being a nibble machine, while the VP could do 8b and 16b ops; the rest was a collection of other improvements.