General performance guidelines?

Willing to start DirectCompute for game development, but I cant find any resources about general usage guidelines, and more importantly general do-and-don’t in matters of performance! Where should i look?

We had some material on DirectCompute optimization at GDC:
http://www.nvidia.com/content/PDF/GDC2011/Nathan_Hoobler.pdf

In general, a lot of the advice from the CUDA programming guide applies to DirectCompute as well.