| |
Cruzer
Registered: Dec 2001
Posts: 1048 |
Clever Routines
Post your unreleased clever routines here...
|
|
... 124 posts hidden. Click here to view all posts....
|
| |
Frantic
Registered: Mar 2003
Posts: 1648 |
@icon: Any sane url is fine for me. codebase sounds good! Thanks!
I actually renamed the site now to "C64 codebase wiki". I think code64 was a bit lame. Everything for the better, I guess. :)
@cruzer: thanks! Yes, the built in asm highlighting is a bit fucked I guess. Better just use <code> instead of <code asm>. That gives straight code without highlighting. I might fix that. Will see if I can patch the editor too so TAB don't jump out of the component, but rather inserts a tab char. Nice if editing the code I guess. But.. All such minor details are a bit in the future right now. Will mostly focus on contents first I guess. |
| |
Durandal
Registered: May 2006
Posts: 30 |
GREAT! I've added a little routine too.
I think the folded plugin could be usefull in the future if the index page grows too big. |
| |
tlr
Registered: Sep 2003
Posts: 1790 |
@frantic: Looks great, good job! |
| |
Scout
Registered: Dec 2002
Posts: 1570 |
I love it!
Thanx.
I hope there will be some moderation because it's just a matter of time when inc $d020 jmp *-3 stuff will be added...
|
| |
Scout
Registered: Dec 2002
Posts: 1570 |
Quote: I love it!
Thanx.
I hope there will be some moderation because it's just a matter of time when inc $d020 jmp *-3 stuff will be added...
And besides the moderation issue, if you look at for instance the fli-fpp-scroller sourcecode (http://wikholm.dyndns.org/~cswiki/doku.php?id=fli-fpp-scroller) you can see that some binary includes are used.
These are missing which makes the example useless because it ain't gonna work this way.
Ofcourse, somebody with coding experience will get this to work but we're all lazy. Just admit it :)
What I'm trying to say here is that adding articles is more than copy/paste into the wiki alone. |
| |
Jayce
Account closed
Registered: Dec 2001
Posts: 39 |
Indeed, to make this work the additional data files are almost obligatory to make some pieces of code actually work :) |
| |
Oswald
Registered: Apr 2002
Posts: 5094 |
Frantic, I think only registered users should be able to edit, so we know who did that. In this case we could tell him to upload the binaries that are missing. |
| |
Luke
Registered: Dec 2004
Posts: 19 |
Quote: On Tour is my contrbution to where you can find a dysp using that technique..
I saw that one...
Sry man, but you don't understand :)
In your production you have no free cycles.
For explain example:
sei
lda #$35
sta $01
lda #$02
sta $fffe
lda #$dc
sta $ffff ;$dc02 as irq vector
lda #$4c
sta $dc02
lda #$00
sta $dc03 ; jmp $xx00
lda #62
sta $dc04
lda #0
sta $dc05
lda #$7f
sta $dc0d
lda $dc0d
lda #$81
sta $dc0d
lda #$0b
sta $d011
bit $d011
bpl *-3
lda #$01
sta $dc0e
lda $dc0d
cli
(routine)
...and now here from $2c00-$3900 per $100 memory bytes
somth like this:
*=$3900
nop
nop
bit $dc0d
dec $d020
inc $d020
rti
*=$3800
bit $ea
bit $dc0d
dec $d020
inc $d020
rti
....
....
....
....
*=2x00 (last one)
nop
dec $d020
inc $d020
bit $dc0d
rti
This is only idea, first need to synchronize irq with
39th column of chars.
It working, but need a NMI irq to stop it in $yy raster line, bcoz timer is continous.
After calculating with 2 sprites takes near 15-23 free
cycles out of irq routine per rasterrline.
|
| |
Oswald
Registered: Apr 2002
Posts: 5094 |
platine, this was already done by ninja, and maybe by others too, you can see it for example here: Real the every 2nd line fli screens (tunel and julia) uses this trick to leave cycles free :) |
| |
JackAsser
Registered: Jun 2002
Posts: 2014 |
I use a similar technique, however a bit developed in The Wild Bunch 4x4 tunnel part with sprite multiplexing. (Yes HCL, I know the multiplexer is a bit bad but let's not blame the actual technique, just blame my lazyness)
Problem with using this technique together with sprites and FLI-generation is that you must stabalize BEFORE the sprite fetches, but has to do the FLI AFTER the sprite fetches. Thus you have a variable amount of cycles between stabalization point and the $d011 write. However, since you're multiplexing and since it's only a matter every 4th line, the multiplexer can easily setup a table with 50 entiries telling the raster code how much to stall each FLI-line.
And since I always had 3x2 sprites I knew that I only had 0,2,4,6 or 8 sprites per line => less different timing routines. A somewhat bigger concern was that I used NMI toghether with IRQs. Timer IRQs are triggered differently using old and new CIA, which is a tiny problem. NMIs interrupting the IRQs setup phase however is a slightly bigger problem since the timing in VICE is b0rked in comparison to the real thing. So all in all I had to had quite many different timing routines to cover all cases.
|
| Previous - 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 - Next |
