[sldev] Reformatting Textures for the cache

[Jason Giglio]

Tofu Linden wrote:
> There is probably a happier medium, but a first point of call would
> be to investigate improvements to the empirically-non-stellar hit-rate
> of the first-level caching before masking or propagating its weaknesses
> through a few more level of experimental caching. :)

I agree 100%.  My priorities right now are understanding the code enough 
so I can add some hit/miss tracking tools, and then implementing ARC and 
  maybe refactoring the system a little.

-Jason