You know, I still can't figure out why that is the case. Why would the ATI line do so much better than the nVidia line? I mean the GF8 series supports 10-bit per component output, so what is it that it is so slow at?
Because the RV670 is simply faster in general computing (GPGPU) applications.
It has to do with the underlying architecture and both R600 and RV670 is based on Very Long Instruction Word (VLIW).