Originally posted by StealthHawk
I'm not understanding how you reached that conclusion. Perhaps you could explain it in more detail? What exactly constitutes a "fully PS2 compliant shader?"
My understanding is that as long as you run all the calls a DX9 shader asks for, it is still compliant. nvidia obviously lowered precision, and rewrote the shader code. ATI is not doing less work, as you seem to imply in another thread. They are doing the exact same amount of work they are supposed to do, with the exact same output.
Now, the question is whether this is cheating. In a benchmark, it could be. Just because your claim may be correct(that ATI is cheating), doesn't mean that you reasoning is(they are somehow not running PS2 compliant shaders).
ok i try to explain, but its just my opinion.
if ati shuttle shader instructions, the shader may not work properly in siturations other than sky and water in GT4, the shader have been altered, they are not strict PS2 shaders that work in every situration, the reason why i think it does not work in all situations is that ati remove it from next driver revision, i just cant believe why ati would not incorporate this "valid optimization" in its PS2 shader instead of just use them in GT4, it just doesnt make sense to me, if this optimization only works in GT4, they are not fully PS2 compliant shaders, they are just PS2 precison water shader and PS2 precison sky shader or PS2 precison GT4 shaders and i am sure futuremark is measuring strict PS2 shader performance that works in every case, not measuring a particular water shader performance, not measuring a particular sky shader performance, not measuring a particular PS2 variant shader performance. futuremark is just measuring strict PS2 shader performance that works in all cases. well of course i have assumed this "shuffle instruction" does not work in all situration. they are just sky shader and water shader with PS2 visual quality in my eyes.