Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!zaphod.mps.ohio-state.edu!van-bc!rsoft!mindlink!a218
From: Charlie_Gibbs@mindlink.UUCP (Charlie Gibbs)
Newsgroups: comp.sys.amiga.emulations
Subject: Re: Emulator Mechanics (sorry long post)
Message-ID: <5008@mindlink.UUCP>
Date: 5 Mar 91 21:29:11 GMT
Organization: MIND LINK! - British Columbia, Canada
Lines: 27

In article <1991Mar6.010141.5905@mintaka.lcs.mit.edu>
rjc@pogo.ai.mit.edu (Ray Cromwell) describes a 6502 emulator that
looks a lot like my SimCPM.  You're right that it would be better
to use MOVE from SR instead of GetCC() to get the flags.  Before
anybody unholsters their flame throwers, I would hasten to point
out that normally I wouldn't think of doing such a dastardly thing,
but in this case it's too much of a performance hit to do otherwise.
We have to keep that loop as tight as possible if we're going to
get any speed at all.

     As in your example, SimCPM's individual instruction emulation
routines fetch the flags so that I can use custom-tailored conversion
routines for more speed.  One trick I use is to use the entire flag
byte as an index into a 256-byte table form which I pull the
translated flag values, again trading memory for speed.

     This makes for lots of places where I need to get the flags.
What I've done in SimCPM is to define a macro that generates either
MOVE from SR or MOVE from CCR depending on the setting of an
assembly-time switch.  The result is two versions of SimCPM - one
runs only on the 68000, while the other runs only on 68010 and
higher processors.  I can't think of any other way to do it that
doesn't itself involve a prohibitive amount of overhead or lots of
self-modifying code.

Charlie_Gibbs@mindlink.UUCP
"I'm cursed with hair from HELL!"  -- Night Court