Comments on: Compiled Bitmaps

by: Dave DeLong

Wed, 22 Jun 2005 20:17:09 +0000

Once again, a fantastically interesting read. Thank you for posting. =D

by: Jonathan Paisley

Thu, 23 Jun 2005 00:48:14 +0000

I’d heard of compiled bitmaps for blitting partially transparent images (with binary per-pixel alpha — so no blending). I wonder what the tradeoffs would be in that case. The compiled blitter can omit writing transparent pixels altogether, whereas other blitters (ignoring graphics hardware acceleration) would have to read the entire source bitmap.

by: corbin

Thu, 23 Jun 2005 08:34:32 +0000

good job, peter! cool article

by: John C. Randolph

Sun, 26 Jun 2005 03:08:32 +0000

Wow, this is a real trip down memory lane. Are you going to cover color-register animation in your next post? :D

I’ve come to realize that some of the techniques I’ve forgotten over the years really are best forgotten...

-jcr

by: Samh

Tue, 23 Aug 2005 07:17:10 +0000

The fastest way to blit would be to use the PowerPCs FPU to copy 64bit values. Trust me I’ve been using it for years.

by: Samh

Wed, 24 Aug 2005 05:32:15 +0000

ok Altivec ... if you’ve got it

by: Anonymous

Fri, 20 Jan 2006 21:02:12 +0000

Hmm... yeah, but on x86 there’s also the string instructions, too...

by: Uli Kusterer

Thu, 30 Mar 2006 04:32:38 +0000

Hi,

just curious: Do you have an ICBM yet? I’d really be interested in hearing whether it works better for them.

Cheers,
— Uli

by: Uli Kusterer

Thu, 30 Mar 2006 04:35:40 +0000

Oh, and to Mr Paisley: I’d guess that you’d get about the same results as RLEing your data would give you... If you know a row is transparent, then you can just skip it. SO I’d doubt it’s an optimisation specific to compiled bitmaps.

by: Paul

Thu, 18 May 2006 07:30:57 +0000

blitting to the screen may be so faster because of a write-combining capability ...

There is a nice S/N here :-) !