18 Jan
2012
18 Jan
'12
1:58 a.m.
Glen Slick wrote:
On Mon, Jan 16,
2012 at 5:06 PM, Jerome H. Fine <jhfinedp3k at compsys.to> wrote:
Can you describe the fine details which make the
QED993 different
from the J11 chip? For example, the early PDP-11 hardware did
not support the MFPT instruction. The PDP-11/73, PDP-11/83 and
PDP-11/84 which supported PMI memory could be tested for. The
Maintenance Register at 177750 had difference values for systems
which included the J11 chip.
Any specific differences for the QED993? Also, what is the official
spelling of QED993? If it is going to be noted, it is best to use the
official spelling used by QED.
QED is "Quickware Engineering & Design, Inc." In the documenation I
have it is "QED 993", with a space, not a dash.
I think one of the big differences is that the QED 993
had no support
for FP11 floating-point instructions at all. It wasn't like a DCJ11
chip without a companion FPJ11 Floating-Point Accelerator (FPA) chip.
It had no support for FP11 floating-point instructions at all. That
seemed to be one of the issues trying to get 2.11BSD to run on the QED
993. When I looked at the 2.11BSD code there seemed to be assumptions
in some places that FP11 instructions were always implemented.
As far as I know, every DEC J11 chip includes the floating point
instructions.
Most of the boards also allow the extra FPA chip, but that just speeds up
the execution of floating point calculations.
Having no floating point instruction support would be a way to distinguish
the QED 993 from a DEC J11 chip if every other implementation (including
DEC, Mentec, emulators, etc.) supports floating point instructions.
One thing about Ersatz-11 is that it would probably be able to configure
the CPU to match a QED 993 since it is so fine grained.
Jerome Fine