Subject: [stella] Give me one more cycle|
From: "Gonzalo Fernández" <horcas@xxxxxxxxxxxxxxxxxx>
Date: Tue, 6 Jan 2004 03:15:58 -0300 (ART)
Hello everybody! Too many time since my last post, so excuse me if this message is too large. Things was very different when I joined the List and my game had chance to be an attractive project, but today amazing things are happening to the 2600. Video, audio in the PCM style, multiple-multiple sprites, against-the-federal-law amount of RAM, 24 bits sprites (from Manuel the great)... so, nothing new can I offer with my demo. The only thing I can do is to work hard on the gameplay, graphics, nice colors... I knew that I had to hurry up (but thinking it better, I started this just for fun). Well, I think the kernel almost is done (aleluya!). Just a few timming problems still are. You may not see too much difference from the last binary I have posted, but I changed all the kernel code and now is cleaner and not like one made by a newbie (wich I am, by the way). No more sections in the kernel (like River Ride), just one smart loop. The funny part that amost every cycle in the scanline is used, and when I tried to get different types of islands I was running out of 1 (ONE!) cycle. Previously a made a lot of optimizations (oh, I worked hard..) but not enough. The last one was the use of an illegal opcode: ATX, wich saved me 2 cycles. I can't belive how useful could be that op. It replaced my old AND # - TAX (4 cycles). Finally, this allowed me to have tree types of islands (Big single and double, and small double). Questions must be made: Do you thing speed is too high? Does the islands shape suck? I downloaded DASM Version 2.20.07 from Andrew's site, why doesn't support illegal opcodes? (I tried ATX, DOP, TOP but in the end had to use the .byte #$XX syntax) Thanks for your attention. -- Gonzalo Fernández.
Description: Binary data