can someone simplify this for us dummies? thank you lol
Mark Cerny talked about cache coherency on the GPU. What it means, that let say you're rendering a lower polygon modal and or lower resolution texture, and bring in a larger polygon model and or larger resolution texture, you don't have to clear the entire GPU caches (this is the norm in GPU architectures), but just clear a section of the cache that had the lower polygon model and or texture.
This saves on loads from main memory, which can take 100's of cycles, and it saves a lot of energy as memory accesses are very power intensive.