So the title of this thread is wrong based on the incorrect 2nd hand statement you got. It should be: Truform on Radeon 9000 - no. (meaning not in hardware and therefore mostly useless)
One of the main ideas behind TruForm is to keep CPU load and AGP traffic low. Doing TruForm in software means you will flood your AGP and host memory interface with triangles. So it is not only a question of CPU power.
There is no such thing as extra CPU power. Putting more work on the CPU is putting more work on the CPU, plain and simple. It doesn't matter if there are some conditions where the final framerate isn't affected, because those conditions quickly change in 3D applications.
I remember speculating here that they removed the FF TnL unit a while before the launch. Made sense to me at the time, I guess. From what I know, the major hardware differences between R200 and RV250/RV250M are: No Hierarchial-Z algorithm usage, no HOS hardware support. Single TCL pipeline. One texture pipeline for 6 texture texturization. Internal cache increased from 2K to 4K.
SirXcalibur , it just seemed strange no mention was made of it in any of thebrochures or reviews i've read of the 9000 . In fact the only mention , I've seen was saying it wasn't supported by HardOCP.
I maybe being a dummy here , but the way I read it was that truform had basically become a sub-set of the displacement mapping.
Could it be that the software part of the Truform is in actual fact the calculation of a truform equivelant displacement map?
Given that Truform is not a hugely supported feature due to it's ATI only nature, it would make sense to remove the fixed function part of this replaceing it with the more programmable displacement mapping unit.
