Is it like 1 SP of G92 @ 1700 Mhz = 1 SP of G94 @ 1700 Mhz....if u think its equal u r wrong
Nope ,, its like 64 SP of G94 @ 1700 MHz < 112/128 SP of G92 @ 1700 MHz and both > 96 SP of G80 @ 1400 MHz

.. particularly under dx9.0c ..
Also 1 more question why do u think thse G92/G94 being based on DX10 unified shader model ...still have better performance in DX 9.0c apps
Is this really a question to ask ?? eeh ?? Did shader alone multiply the performance ?? Then going from 24 ps + 8 vs to 96 SP (say 3*24 ps + 3*8 vs) unified should actually give u 3 times the performance wouldn't it ?? I don't see the 8800GTS 320 MB/640 MB scoring 3 times in 3DMark , or any other game for that instance , compared to a 7900GTX !!!
Its about Core Clock , Mem. Clocks , ROP's , Texture Units , independent Shader Processing Units , Mem. Interface , No. of SP's etc. etc. ... combine all that and u get just over 2 times the performance compared between a standard 7900GTX and a 8800GTS (suppose) ... and the two cards are generations apart

..
Is it really hard to accept that dx9.0c can't properly utilize all the features offered by current gen. of dx10 cards (unified shader architecture , geometry shaders etc. etc.) ?? The performance charts are still based upon games which are dx9.0c based , and extended to dx10 (the other way round is not possible if i'm not mistaken) ...