For starters, it should be mentioned that, by default, the use of DST (and thus PCF) is enabled in the benchmark provided the hardware supports it.
And this is basically where the problem lies - DST and PCF are currently only supported by NVIDIA's range of video cards (from the GeForce 3 and upward). There are currently rumours that another IHV will eventually have DirectX 9 capable hardware with support for DST, but not for the PCF component.
To compound this problem, the patents and intellectual property rights for hardware Depth Stencil Texture acceleration are owned by... That's right... NVIDIA. This means that other IHVs that wish to utilise this functionality in their own hardware will most likely end up having to pay a hefty fee.
This also tends to go against one of FutureMark's own rules - Not to use IHV specific features, or indeed features that are only supported by a single IHV. Of course, part of this argument can be negated by bringing up the aforementioned rumours of another IHV supporting this feature in the future.
Read it your self
http://www.elitebastards.com/page.php?pageid=10599&head=1&comments=1
Performance drop with DST disabled:
Video card: DST off:
GeForce 6800 Ultra -10.5%
GeForce 6800GT -10.3%
Radeon X800 XT PE 0.4%
Radeon X800 XT 0.7%
As you can see, both NV40-based boards lose around 10% of their overall 3DMark score at default settings once DST is turned off, whereas the two ATI boards remain basically untouched by its removal. Indeed, with DST off, the tables are very much turned in ATIs favour, from a tie between the high-end 6800 Ultra and X800 XT Platinum Edition at pure default settings to a margin of victory of several hundred points to the ATI card with DST switched off. Indeed, even the non-Platinum Edition X800 XT squeezes a few hundred points ahead of the 6800 Ultra.
