Already covered in post above... and I found my old post on this:
http://forums.anandtech.com/showpost.php?p=37669975&postcount=14
Basically HWS is present in Fiji already. It's an improved 2x ACEs => 1x HWS unit that can perform the roles of 2x ACEs as well as new enhanced functions.
If you guys are looking for enhance DX11 function, it's this:
Refer to the wave instruction cache/buffer.
AMD's GCN currently lack a deep buffer so the wavefront instructions have to constantly come in to keep the shaders running. If the CPU becomes bottlenecked (in DX11, single threaded renderin), it stalls.
This change should result in improved DX11 performance in those scenarios of either:
1. Weak CPU or poor single thread IPC CPU in demanding games.
2. Games that hammer a single thread, such as PhysX heavy titles running on AMD DX11.
3. Improve minimum FPS in DX11 IF the bottleneck is the CPU.