Re: [stella] perfect timing

Subject: Re: [stella] perfect timing
From: Thomas Jentzsch <tjentzsch@xxxxxx>
Date: Sun, 11 Nov 2001 11:14:34 +0100
Glenn Saunders wrote:
> Okay, I think I managed to get a 2 line kernel working completely without 
> the use of WSYNCs at all!

> Boy was it a lot of work.

I know! You have been warned :-)

> Thomas, I integrated your new Tombstone indexer routines which seem to work 
> but the first row of tombstones only shows 7 lines instead of 8.  Do you 
> what that might be?

Yes, I think Y starts one to low, so the first block is one line to small.
I don't have your actual source, but if you are using mine, then it
should work, if you simply swap DEY, TYA. That way, A get's the previous
Y value, which in result will delay your row-blocks by one line.

> So right now we've got 26 cycles of processing time before I blank out the
> playfield for the 2nd line of the linepair (the blankout code can move 
> around some of course) and then a whopping 64 cycles in a row.

> So that's a total of 90 cycles free for the line-pair.

90 cycles sound's a bit to low for me. How are the other 62 used? (PF +
HMOVE + loop = max. 55)

> It takes 30 cycles a piece, by my calculations, to do the missiles.  So 
> that's 60 cycles right there.  That would consume the majority of the 
> sprite line, leaving barely enough to generate one car.

> It looks to me like there is enough time to do 3 of the 4 objects, but not 
> all 4.

With 26 cycles (which is my count), you'd gain 8 more cycles here, the
optimal player drawing routine need 18 cycles, so 38 cycles are enough.

Have fun!
Thomas Jentzsch         | *** Every bit is sacred ! ***
tjentzsch at web dot de |

Archives (includes files) at
Unsub & more at

Current Thread