I get 150 fps with the C code on the fire example now… That is twice as
fast as PTC with the old conversion routines… (actually, the 8 to 16
bit C routine is as fast as the assembler routine now)
Tell me what you think… Personally, I will always have the latest Ariel
version within the PTC x11 distribution so that people don’t have to
Looks very nice.
I re-organized the SDL blitting routines to be much lower overhead,
and they now use the same structure for arguments that your assembly
Since our C routines are very similar, I’ll just stick to the ones
already in SDL, and use your assembly routines directly.
My 8-bit --> N-bit lookup table code seems as fast as your assembly,
so I’m using the x86 and mmx routines for 32-bit conversion.
This means that SDL now has x86 and MMX capability!
Thank you very much.
If the calling interface doesn’t change, anyone should be able to drop
in new versions of the mmx_32.asm and x86_32.asm files, and get instant
speedup. Also, it’s trivial to add more specialized C and asm blit
routines as people desire them.
The unaccelerated colorkey and alpha blending blitters are still dog