Actually im calling out the Ati troll that takes any part of an article that he thinks is beneficial to his argument and beats it to death in every single video thread he comes across.
Since you seem to think im trolling we can go into detail.
You have effect X and effect Y
ATis buffer for these effects is 16 pixels (4x4)
Nvs buffer for these effects is 1024 pixels (64x16)
Now, lets say we got with beyond 3ds worst case scenario, and say that the particular effect we are putting together is relatively small X can be 10x10, and Y can be 10x10.
But, we can make it even worse for nvidia and make X overlap 4 Ys, so the calculation must be repeated 4 times.
So we have to calculate X effect + Y effect = output 4 times individcually on nv hardware.
On ATi without going into insane logistics and just giving a best case scenario, lets assume they perfectly overlap.
ATi would have to make 9 combination calculations on the exact same effect in an ideal situation.
So while the buffer is smaller individually, more calculations have to be made. Its a benefit and a drawback at the same time.
Potentially by atis design more effects could be combined in total on the same size total buffer, but by nvidias design less total calculations are made.
Facebook
Twitter