1. Pixel optim
I have managed to put together a graphics library, however, the routines
I'm using are very slow (go figure why :)) Anyway, do anyone know a
optimization for this code?:
global procedure display_emb_vscreen(atom xoffset, atom yoffset, atom
xparent, atom source)
atom cache1, cache2, xvar, yvar, color, readpos
cache1 = peek(source)
cache2 = peek(source+1)
xvar = cache1*256+cache2
cache1 = peek(source+2)
cache2 = peek(source+3)
yvar = cache1*256+cache2
readpos = 4
for y = 1 to yvar do
for x = 1 to xvar do
color = peek(source+readpos)
if color != xparent then
vscreen[y+yoffset][x+xoffset] = color
end if
readpos = readpos + 1
end for
end for
end procedure
xvar and yvar is the size of the picture :) source is the hex address of
where it starts
BTW: Sorry for stealing the idea of a virtual screen :) but this routine is
20% faster than my old
2. Re: Pixel optim
>From: Alexander <Carnivals at INAME.COM>
>Anyway, do anyone know a
>optimization for this code?:
Just a very basic transcopy routine. doesn't do clipping or much of
anything..
procedure transcopy(integer x, integer y, integer width, integer height,
integer trans, atom src, atom dest)
integer src_offset
integer pix
src_offset = 0
dest += (y*SCREEN_WIDTH)+x
for i = 1 to height do
for j = 0 to width - 1 do
pix = peek(src + src_offset)
if pix!=trans then
poke(dest + j, pix)
end if
src_offset += 1
end for
dest += SCREEN_WIDTH
end for
end procedure
but if you really want speed the you might want to do this in assembly, or
check out some of the other graphics libraries available.
_____________________________________________________________________________________
Get more from the Web. FREE MSN Explorer download : http://explorer.msn.com
3. Re: Pixel optim
On Sun, 26 Nov 2000, mic _ wrote:
> >From: Alexander <Carnivals at INAME.COM>
> >Anyway, do anyone know a
> >optimization for this code?:
>
>
> Just a very basic transcopy routine. doesn't do clipping or much of
> anything..
>
>
> procedure transcopy(integer x, integer y, integer width, integer height,
> integer trans, atom src, atom dest)
> integer src_offset
> integer pix
>
> src_offset = 0
> dest += (y*SCREEN_WIDTH)+x
>
> for i = 1 to height do
> for j = 0 to width - 1 do
> pix = peek(src + src_offset)
> if pix!=trans then
> poke(dest + j, pix)
> end if
> src_offset += 1
> end for
> dest += SCREEN_WIDTH
> end for
>
> end procedure
>
>
> but if you really want speed the you might want to do this in assembly, or
> check out some of the other graphics libraries available.
>
>
> _____________________________________________________________________________________
> Get more from the Web. FREE MSN Explorer download : http://explorer.msn.com
>
It's even faster if you load the memory to a sequence and operate on that
- using peek and poke for the whole image is much faster than peeking and
poking individual bytes.
Jeff
