2 Sep
2008
2 Sep
'08
7:46 a.m.
On Mon, Sep 01, 2008 at 11:45:00PM -0700, Eric Smith wrote:
Ethan Dicks wrote:
I suppose there could be a code advantage in splitting things up like
that, but I don't see much of a in-core storage advantage, unless you
pack, for example, 3 variables in 4 machine words (in which case, the
code could get fairly messy).
Fundamentally, I think one would have to scale up
all the 16-bit Z-pointers
to 24-bits internally to a theoretical PDP-8 Z-machine, and there's a lot
of 16-bit math going on under the hood.
That sounds like the really hard way to write an interpreter for the
PDP-8. It would be much easier if you just pretend it's an 8-bit
machine, and do 8-bit arithmetic, and 16-bit double precision, just as
was done in the 6502 interpreter. A fairly naive implementation should suffice for most
of the early V1-3
games. Once those are running, add swapping of impure data, and you'll
be able to run larger games.
Sure... adding impure swap would make the larger games fit, at what might
be a hefty performance penalty - but it should work.
-ethan
--
Ethan Dicks, A-333-S Current South Pole Weather at 2-Sep-2008 at 14:40 Z
South Pole Station
PSC 468 Box 400 Temp -54.4 F (-48.0 C) Windchill -93.5 F (-69.7 C)
APO AP 96598 Wind 17.5 kts Grid 32 Barometer 680.5 mb (10609 ft)
Ethan.Dicks at usap.gov http://penguincentral.com/penguincentral.html