Imagination PowerVR SDK Blog

uniform caching


#1

Dear Team



what is the motivation behind the cache that you use in the deferred rending demo ?

CPVRTMap<int, int> *m_pUniformMapping;

should it be handled by the TOOLS ( if i remember , you were doing the same caching technique in the POWERVR Engine).



if it is part of best practice on performance , why it is not use everywhere else ?

:slight_smile:



regards

david


#2

Hi David,



It’s a good practice to cache values in your app to keep calls into GL to a minimum. Although the Get operations are very cheap, not doing them at all is even better :slight_smile:



The DeferredRendering demo used this technique as it has to manage multiple effects. In other demos (e.g. the Water demo), the values are cached, but a map hasn’t been used.

You’re right that we should use techniques like this consistently. When we move to an updated tools framework in the future, it will be a good opportunity to clean up a bit.



Thanks,

Joe