[Coco] No more dream ports (Was:Wolfenstein 3-D on a 4.77mhz 8088? So, 6809 port, anyone?)

[Steve Bjork]

On 4/24/2013 7:33 PM, Luis Antoniosi (CoCoDemus) wrote:
> Of course I was talking to make from scratch and not port, also C on coco
> sucks.
>
> I do agree with all hardware limitations and I never dreamed about a
> wolf-3d clone but some sort of 3d game with some texture mapping.
>
> But one thing we have now: new tools that allows us to squeeze every cpu
> cycle on those CPUs and emulators to debug and test it line by line. just
> look at the C64 demo scene to see how far they went. but of course, as I
> said demo is not a game. a demo is all controlled and cycle counted.
>
What do you mean by the new tools that we have now?

When I was coding for the CoCo, I had everything I needed to squeeze 
every CPU cycle of the CoCo.  (Remember, 25% of my coding time was used 
for creating the tools used in my coding.)  I see nothing that came out 
since I work the CoCo to make for faster code.  Yes, we have learned a 
few new technique in writing code or using the hardware.  But still, 
these technique only give us a few percent of game improvement.

I never had the need for emulators or debuggers.  My code had a 
developer OS built in.  This would give me the timing/resource 
information on very code/graphic object in the game.  (I would remove 
the OS when creating the final version for sell.)  As for debugging, I 
only needed to look at the source code or the timing/resource 
information to know what was going and how to fix.

As for squeeze every CPU cycle, I knew the timing cycle and byte count 
of every instruction that I wrote.  Little things like using the LDY was 
really the LDX instruction with a Pre-Byte header before it.  So, I try 
not to use the Y register since loading took an extra clock and extra 
byte.  (It should be noted that LEAX and LEAY was the same speed and size.)

Back in those days, we coded "down to the metal."  We knew just how the 
6809 and the CoCo's hardware work.  This is something that's lost on 
modern coders.  Once a year, I still give a talk at UCLA to the new 
programmers for one of my old professors. (Well, his replacement since 
he retired a decade ago.) My two hour talk covers how to "know and own 
your code" and cut out the waste.  After all, wasteful code cost money.

Yes, we will learn new ways to do things.  But none of the new technique 
will make CoCo any newer than a 30 year old computer.

Steve