NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/comp/sys/m6809/591

Internet Message Format | 1993-01-27 | 3.3 KB

Path: sparky!uunet!news.univie.ac.at!scsing.switch.ch!univ-lyon1.fr!ghost.dsi.unimi.it!rpi!gatech!usenet.ins.cwru.edu!agate!doc.ic.ac.uk!uknet!comlab.ox.ac.uk!newshost!rpb
From: [e-mail] (Ray Bellis)
Newsgroups: comp.sys.m6809
Subject: PD m6809 simulator to be available soon
Message-ID: <[e-mail]>
Date: 27 Jan 93 12:51:33 GMT
References: <[e-mail]>
    <[e-mail]><[e-mail]>
    <[e-mail]> <[e-mail]>
Organization: Dept. of Experimental Psychology, Oxford, England.
Lines: 62
In-reply-to: [e-mail]'s message of 26 Jan 93 06:15:49 GMT


In article <[e-mail]> [e-mail] (Ben - Williams) writes:

   That's not too bad a speed! You do have a bit of horsepower, though. :^)

I'm not too worried about coding efficiency, I'll be running it on a
150 MHz Alpha very shortly :-)

   I can get almost realtime for the loop you show in the NON debug version
   of my emulator; about 900k/sec, same instructions. That's on a 25 mhz 68040,
   all caches on.

   My code is a great deal simpler than yours, though, and involves no
   function calls to anything, anywhere, other than one call to execute the
   instruction itself. All the non-indexed modes are coded completely inline,
   your example being an immediate mode, it falls in that group.

My example wasn't immediate mode.  The `fetch_operand' function call
handles operands for all modes except inherent.  My `execute' function
first of all decodes the addressing mode and then uses a huge `switch'
to find the function that implements the instruction.  I suspect that
a lot of my CPU time is `wasted' by the C++ implememtation, and by the
virtual functions that implement read/write `memory cycles'.

   Emulators are a kick, eh? <grin>

Yeah, I got quite a kick the first time I put that simple 2 byte
counter into the system and traced the registers being incremented
etc.

------------
Announcement:
------------

My C++ source code emulator of the MC6809 will be available to all
and sundry very shortly.  If people are interested in Beta testing,
please contact me, as I'm not sure if my interpretation of how things
like the `C' flag is set are correct, so I'd really like people to
verify the accuracy of my simulation.  At the moment, the only un-
implemented facilities are asynchronous interrupts, as I haven't
figured out a good mechanism for handling these yet.

I have a simple `machine' derived from the mc6809 base class which
provides a read/write port at address 0xc000 and either gets or puts
a character to stdin/stdout.  This derived class is written using
POSIX `tc' functions to put the tty into `raw' mode but could easily
be rewritten to use DOS `conio' functions.

Now all I've got to do is write an operating system, as the emulator
was orignally just a tool for developing the OS for a simple MC6809
based controller that I'm designing...

Ray.

--
------------------------------------------------------------------------------
R. P. Bellis                E-Mail:    <[e-mail]>
Dept. of Experimental Psychology    Whois:    (RB83)
University of Oxford            Tel:    [phone]
South Parks Road            Fax:    [phone]
Oxford OX1 3UD
------------------------------------------------------------------------------