Douglas Quebbeman wrote:
IIRC, the Eclipse architecture was an evolution of the
DG Nova architecture.
It's development is the subject of Tracy Kidder's "Soul of a New
Machine",
a book I'd highly recommend reading.
Sorta kinda. The Eagle architecture has some commonality with the
Nova and 16-bit Eclipse family, and borrows on the technique of recycling
meaningless instructions from the Nova in order to create instruction set
space. Prior to loading microcode an Eagle is a brain-damaged unmapped
Nova 800 -- more or less all of the Nova instruction set, minus the page zero
autoincrement/autodecrement instructions. Enough that you can boot off disk
and run enough code to find the microcode file and get on with life.
The memory architecture is utterly differerent, supporting not only
paging (it was about time!) but with eight rings -- which is why we chose it
as the basis for our B-level OS work. While it maintained the ever-so-awful
nova I/O bus the architecture allows support for seven distinct busses
(although I'm not familiar with any machines which had more than two).
One interesting characteristic of the Eagle that bit a lot of code generators
is that many instructions had fields which were defined as "reserved, must
be zero:. On the MV8000 these bits were ignored; on the MV4000 and 10000
having these bits nonzero resulted in traps. Ouch.
Cheers,
Chris
--
Chris Kennedy
chris(a)mainecoon.com
http://www.mainecoon.com
PGP fingerprint: 4E99 10B6 7253 B048 6685 6CBC 55E1 20A3 108D AB97