24 Sep
2001
24 Sep
'01
6:56 a.m.
Mike McFadden wrote:
I'm not the expert on microcoding versus writable
control stores. I seem to
dimely remember that there was a whole section about microcoding in the 1981
book by Tracy Kidder "Soul of a new machine". Lots of microcode was used to
emulate "old" instructions from previous machines. Also to perform complex
series of instructions.
The Eagle (MV) 32-bit members of the Eclipse family, like the 16-bit members
before them, were indeed microcoded -- with the difference that the MVs had
soft microcode that was loaded off of magnetic media (floppies on the original
MV8K; later from files stored in the native AOS/VS file system).
Faced with the issue of _how_ to read the microcode the MV designers
did something modestly clever. Since the MV is an almost-proper superset
of the Nova, and since the Nova ISA was well understood, when a MV first
powers up it thinks it's a slightly brain damaged Nova 800 (there's no
page zero auto inc/auto dec instructions and most "do nothing" I/O an
skip instructions are not honored), which gave them a reasonably
complex abstraction upon which to open and traverse file systems
looking for the MV microcode...
--
Chris Kennedy
chris(a)mainecoon.com
http://www.mainecoon.com
PGP fingerprint: 4E99 10B6 7253 B048 6685 6CBC 55E1 20A3 108D AB97