Pemalite said: Exactly. ;) |
Another piece of tidbit is that shader model 6 goes one step further than SPIR-V in terms of exposing more features from the GCN microachitecture ...
WaveOnce and GlobalOrderedIncrement along with WaveGetOrderedIndex are pretty specific to AMD ... (DPP from GCN3 is also a good idea for implementing quad swizzles)