Is there any OpenGL ES 2.0 level hardware where dependent texture reads/dynamic texture lookup will not incur overhead?
Apple recommends avoiding dynamic texture lookups since it increases the latency of accessing memory when sampling from the texture.
Is this a problem specific to the GPU hardware Apple uses in their devices or is this just a general problem with all OpenGL ES 2.0 level hardware? i.e, how does the Adreno GPU family handle dependent texture reads?