ATI's entry into the hardware T & L arena actually bodes very well for consumers, developers, ATI and even NVIDIA, who's successful charge into T & L last year was previously followed only by S3 featuring it on the box but never getting it to work well enough to turn in on in their drivers, which brings us to the often-overlooked point that none of this stuff is a walk in the park! Getting a new feature into silicon is hard enough. Making it outperform existing methods is the kind of constant challenge that leads to all nighters, cold pizza and divorce.

Other hard work that has gone into the Radeon 256 is for hardware accelerated key frame interpolation, 3D textures, support for all 3 types of Direct-X Bump Mapping and for Environment Mapping and Shadow Mapping.

The Radeon 256 also performs hardware vertex skinning, an interesting if not immediately useful feature that, if well implemented (meaning better than NVIDIA's version), well marketed to OEMs and consumers and successfully evangelized to developers, could become a standard for software to get one step closer to the hardware. At up to four matrices, it is already more precise than NVIDIA's two matrices version but speed is the other side of beauty and speed it still the king around here. Until we can see implementation performance side by side, we won't know if NVIDIA's 2x is good enough or ATI's 4x is fast enough.

Of all the new features, vertex skinning may prove to be the most practical as it utilizes an existing format (reportedly a straight port from 3DStudio Max) to create the desired content but 3D mapping is probably the sexiest and with vocal support from id Software's John Carmack, its chances of catching on (remember OpenGL?) with both developers and other hardware vendors are much improved.