Sunday, July 3, 2016

Day 49: Profiling

It's quite odd to keep looking for optimizations after the big ones roll through. Cell culling and dynamic levels of detail (distance mapping and tessellation) are big ones, but how about the code that lies in between? Why not break when detecting cells for heightmap movement? Why are there still post-increments?

One of the bigger optimizations, though, is the setting of texture resources. Since most shaders only use two textures at a time, the larger terrain shader would only have to set its textures once (3-9, to be exact).

As for capstone? Just some bugfixing here and there; oddly enough, multiple bridges could be used on an insect and that character could aim while in a gravity well. Fixed those up!

What next? I can still see if I can optimize more...

No comments:

Post a Comment