Re: [stella] perfect timing

Subject: Re: [stella] perfect timing
From: Glenn Saunders <cybpunks2@xxxxxxxxxxxxx>
Date: Sun, 11 Nov 2001 15:06:14 -0800
At 10:53 PM 11/11/2001 +0100, you wrote:
You should move the TombstoneIndexer routine at the end of the loop.
That way you save the extra JMP TombstoneScanline (3 cycles).

But the downside of that is you can't do important sprite stores at the end of the scanline. You need to be able to do that to avoid the tearing problems.

That problem can be solved with enabling/disabling VDEL quite easy. The

Even with VDEL I still have to be mindful of where I perform the update to GRPx. I have to do it either before or after the sprite is displayed.

What I'm doing now is building two kernels where both cars are updated either right before the start of graphics for line 1 or right before the start of graphics of line 2. That's why I have the indexing stuff where it is, because otherwise it would eat up all the important cycles at the end of line 2 which I need to prep the sprite stores.

The only time I will need VDEL is if the startposition of either car is offset. I will then need a 3rd kernel that handles the sprites with VDEL in mind (the style where the sprites get manually zeroed out).

With those three kernels I will have more than enough time to wedge the missile code for one missile in there, and then I just need to find a safe spot to store the enable and width writes. The movement register should be safe to rewrite anywhere as long as it's not right after the HMOVE.

I can update just one missile per line-pair and that should do it, at the cost of vertical missile movement resolution.

I know this doesn't sound exactly like what you are suggesting, but given my level of skill with the 2600 right now, it's the best I can actually code for.

---------------------------------------------------------------------------------------------- Archives (includes files) at Unsub & more at

Current Thread