Re: [stella] "kernel panic!"

Subject: Re: [stella] "kernel panic!"
From: Ruffin Bailey <rufbo1@xxxxxxxxxxx>
Date: Thu, 04 Jul 2002 04:19:53 -0400
on 7/3/02 10:09 PM, KirkIsrael@xxxxxxxxxxxxx at KirkIsrael@xxxxxxxxxxxxx
wrote:

>> At 03.07.2002, 01:15, Kirk wrote:
>>> And man, this is for one lonely sprite! Any reccomendations
>>> for cleaning it up?
>> 
>> Search the archives for "skipDraw". :-)
> 
> That's pretty cool, and I might move to some of those more sophisticated
> techniques, but I found a great stopgap for myself, since right now I have
> some RAM to play with:
[snip]

I've gotten to about the same place in the code I'm playing with -- I've got
four things (two players and two missiles) on the screen now, with a
y-positioning byte in RAM for each, and four checks per scan to see which
should be displayed.  Of course the bins that I'm kicking out readily show
that my current [nonexistent] timing makes for some pretty wacky
"pogo-sticking" depending on how far down the scan each object occurs (at
least I *think* that's what's happening).

Thomas' skipDraw message
(http://biglist.com/lists/stella/archives/200110/msg00229.html) is
essentially the same thing Manuel did to my cheesy demo when he stuck it
into the Stellalist Beta Demo program.  Though skipDraw is quick, you
unfortunately still can't get in four checks per scan, much less four checks
per line with writes to the ENAMx's and GRPx's.

So though the skipDraw is fast, I don't know that it fixes Kirk's (or my)
problem.

I've been thinking I'd do the same thing Kirk has proposed, using a byte of
RAM for each object as a buffer [and use branching in the kernel for PF
updates].  LDA from zero page takes 3 cycles and STA into zero page also
takes 3.  That's 6 per object checked, making 24 for four objects.  Since I
can "carry over" one value from the preceding scan line, that's 24-3 = 21.

It would appear from Nick's Cycle Counting paper that this is going to put
me one cycle over where I want to be for one object.

That sound about right?

For JoustPong, however, it seems to this biannual hacker that you should be
right on the money.  Two players, one ball, voila.

Ruffin Bailey

----------------------------------------------------------------------------------------------
Archives (includes files) at http://www.biglist.com/lists/stella/archives/
Unsub & more at http://www.biglist.com/lists/stella/


Current Thread