On 5/26/2016 2:25 PM, Jay West wrote:
Of course, this p-code was executed
interpretively. By missionary instructions;)
On the microcoded machines, the 1600 and Ultimate (honeywell level 6,
and also custom bit slice microengines) systems in particular, the
interpreter was assisted by a macro instruction which essentially
implemented the jump to a table based on the next p instruction pointed
at by the p-code instruction counter.
At the end of the pick assembly code (which was executed by firmware)
the special instruction was executed in the stream of pick assembly
instructions.
Later versions implemented execution of 8 or 10 instructions directly in
the firmware, so sometimes when some versions of the firmware came out
it was not just a branch to code to execute all the p-code, but
sometimes a lot of pcode would be executed by the firmware, eventually
coming up for air and back to some pick firmware when a non firmware
instruction was encountered. A very interesting mix of executing code
at both the macro level and micro level.
for completeness of this, if I've not lost you, there was a machine
built by Pick which was also microcoded, and as far as I know those were
the only microcoded machines for pick.
The above concept worked in varying degrees with the cross compiled
systems which ran on 68000 and other machines as well.
Thanks
jim
