Apple-2-SW/prog8
1
0
Fork 0
mirror of https://github.com/irmen/prog8.git synced 2025-03-08 03:35:29 +00:00
Code Issues Projects Releases Wiki Activity

optimizing gfx2.fill() 4 color mode

Browse Source
This commit is contained in:
Irmen de Jong 2024-08-25 14:48:42 +02:00
parent 03145630f8
commit c4c4dcf2b3
2 changed files with 92 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

111
compiler/res/prog8lib/cx16/gfx2.p8
View File

@ -628,7 +628,16 @@ gfx2 {
}
}
ubyte[4] fill_amask = [$c0,$30,$0c,$03] ; array of cmask bytes
sub fill(uword x, uword y, ubyte new_color) {
; reuse a few virtual registers in ZP for variables
&ubyte fillm = &cx16.r7L
&ubyte seedm = &cx16.r8L
&ubyte cmask = &cx16.r8H
&ubyte vub = &cx16.r13L
&ubyte nvub = &cx16.r13H
; Non-recursive scanline flood fill.
; based loosely on code found here https://www.codeproject.com/Articles/6017/QuickFill-An-efficient-flood-fill-algorithm
; with the fixes applied to the seedfill_4 routine as mentioned in the comments.
@ -708,6 +717,7 @@ gfx2 {
return
if xx<0 or xx>width-1 or yy<0 or yy>height-1
return
if gfx2.active_mode == 2 set_color_masks()
push_stack(xx, xx, yy, 1)
push_stack(xx, xx, yy + 1, -1)
word left = 0
@ -779,30 +789,93 @@ skip:
}
sub fill_scanline_left_2bpp() -> bool {
; TODO optimize this to use vera auto-decrements, but requires masking etc because of 4 pixels per byte...
cx16.r9s = xx
while xx >= 0 {
if pget(xx as uword, yy as uword) as ubyte != cx16.r11L
break
xx--
uword vx = xx as uword
void gfx2.addr_mul_24_for_highres_4c(yy as uword,vx)
cx16.r1L |= %0001_1000 ; auto decrement
set_vera_address()
cmask = fill_amask[lsb(vx) & 3] ; set the color mask for the first color pel
repeat {
vub = cx16.VERA_DATA0 ; read the VERA color data for 4 pixels
if cmask == $03 { ; only speed fill from far right
; speed fill
if vub == seedm { ; all four colors match the seed
nvub = fillm ; replace all four colors at once
xx -= 4
goto set_byte ; go on
}
}
; replace one color at a time
nvub = vub
while cmask != 0 {
if vub & cmask == seedm & cmask {
nvub &= ~cmask
nvub |= cmask & fillm
; %asm{{
; lda p8v_cmask
; trb p8v_nvub
; and p8v_fillm
; tsb p8v_nvub
; }}
xx--
cmask <<= 2
} else { ; not the seed color, finish here
cx16.VERA_DATA1 = nvub
return vx == xx
}
}
set_byte:
cx16.VERA_DATA1 = nvub
if xx <= 0 break
cmask = $03
}
if xx!=cx16.r9s {
horizontal_line(xx+1 as uword, yy as uword, cx16.r9s-xx as uword, cx16.r10L)
return false
}
return true
return vx == xx
}
sub fill_scanline_right_2bpp() {
; TODO optimize this to use vera auto-increments, but requires masking etc because of 4 pixels per byte...
cx16.r9s = xx
while xx <= width-1 {
if pget(xx as uword, yy as uword) as ubyte != cx16.r11L
break
xx++
void gfx2.addr_mul_24_for_highres_4c(yy as uword,xx as uword)
cx16.r1L |= %00010000 ; auto increment
set_vera_address()
cmask = fill_amask[lsb(xx) & 3] ; set the color mask for the first color pel
repeat {
vub = cx16.VERA_DATA0 ; read the VERA color data for 4 pixels
; speed fill
if vub == seedm { ; all four colors match the seed
nvub = fillm ; replace all four colors at once
xx += 4
goto set_byte ; go on
}
; replace one color at a time
nvub = vub
while cmask != 0 {
if vub & cmask == seedm & cmask {
nvub &= ~cmask
nvub |= cmask & fillm
; %asm{{
; lda p8v_cmask
; trb p8v_nvub
; and p8v_fillm
; tsb p8v_nvub
; }}
xx++
cmask >>= 2
} else { ; not the seed color finish here
cx16.VERA_DATA1 = nvub
return
}
}
set_byte:
cx16.VERA_DATA1 = nvub
if xx >= gfx2.width-1 break
cmask = $C0
}
if xx!=cx16.r9s
horizontal_line(cx16.r9, yy as uword, xx-cx16.r9s as uword, cx16.r10L)
}
sub set_color_masks() {
seedm = cx16.r11L | (cx16.r11L<<2) | (cx16.r11L<<4) | (cx16.r11L<<6) ; seed mask
fillm = cx16.r10L | (cx16.r10L<<2) | (cx16.r10L<<4) | (cx16.r10L<<6) ; fill mask
}
}

3
docs/source/todo.rst
View File

@ -1,9 +1,6 @@
TODO
====
Optimize gfx2 fill_scanline_XXX routines also for the 2bpp modes (4c hires)
See open issues on github.
Re-generate the skeletons doc files.