17 Feb
2012
17 Feb
'12
2:04 p.m.
Re:
On Feb 8, 2012, at 6:06 PM, William Maddox wrote:
The Accent OS could switch among multiple microcoded
instruction
sets upon a context switch. Spice Lisp (forerunner of CMU Common
Lisp) had its own Lisp-machine-like instruction set, while the OS
and system utilities ran the Pascal (Q-code) instruction set.
I don't know if any of this ever escaped from CMU, however.
Burroughs B1700 / B1800 / B1900 series did this, too (circa 1973?)
The B1700 ran microcode via a "pico code" interpreter, and each
program could have its own instruction set. IIRC, there were two
COBOL instruction sets (two sets of microcode): one for programs using
a small number of variables (hence, fewer address bits needed), and one
for large numbers of variables. Plus, different microcodes for the
FORTRAN and ALGOL, perhaps others.
Stan