Will there be a HVC miner optimization for non-GCN cards, ie. HD 5850, 5870, 6950, 6970, etc? They run slow, barely 4 to 4.5 mhs.
For HVC, probably not, unless AMD fixes their compiler for VLIW, which is unlikely.
Looks like AMD isn't focusing on improving VLIW any time soon. Mantle and the open API is their focus with the GCN technology.