(mods: not sure if this belongs here or in General Talk, feel free to move it. Just figured it would be of interest to a lot of projects that do 32-bit surfaces etc)
My first optimization in ARM asm on GP2X h34r:
http://www.scoopex1988.org/article-gp2x-ar...6-bpp-blit.html
Derived from Senor Quack's (ie. A_SN's) version, it uses 8 instructions per pixel-pair instead of 12, and one big burst read from memory per iteration instead of two.
Version 2 will have some small optimizations, this version is for comparison and readability, and direct replacement. I will also make a full speed comparison, but I figure an execution time a little over 67% compared to the one on the wiki.
I'd be happy to collab with lib and engine people!
My first optimization in ARM asm on GP2X h34r:
http://www.scoopex1988.org/article-gp2x-ar...6-bpp-blit.html
Derived from Senor Quack's (ie. A_SN's) version, it uses 8 instructions per pixel-pair instead of 12, and one big burst read from memory per iteration instead of two.
Version 2 will have some small optimizations, this version is for comparison and readability, and direct replacement. I will also make a full speed comparison, but I figure an execution time a little over 67% compared to the one on the wiki.
I'd be happy to collab with lib and engine people!
Last edited by a moderator: