godsic

LOCAL MEMORY PACKING

Discussion created by godsic on Aug 21, 2009

AMD what about packing of GPU local memory, similary to DXT and others in graphics APIs.

Yes it hard because it must be some arithmetic packings.  

If you implement some methods with TMU and on-driver level packing ( during Unmaping for example).

It can minimize memory bottlenecks, memory footprints and therefore increase performance of fetching operations. 

Outcomes