Since the RIVA 128, NVIDIA has been on a strict 6 month product cycle, and generally, this has led to a totally new NVIDIA product in the fall, followed by a "spring refresh," where the original product is updated and released as a new one. Usually, that spring refresh means a few new features and, typically, a higher clock rate.
The RIVA 128 ZX followed the original RIVA 128 by adding support for 8MB of memory and AGP 2X support. We saw the "spring refresh" again with the RIVA TNT2, released earlier this year as the follow-up to the TNT. The TNT2 added AGP 4X support, the ability to handle up to 32MB of RAM, and was built on a cooler running 0.25-micron process.
For the fall of 1999, we have the all-new GeForce 256 from NVIDIA, which makes significant strides forward with hardware transform and lighting and the requisite fillrate increases. Just like previous "next-generation" parts, this is built on an even smaller 0.22-micron process. However, this fall, NVIDIA has also introduced a number of other chips to complement the GeForce, throwing off their theoretical 6 month product cycle ever so slightly. Fortunately, it’s thrown off for the better resulting in even more products. Of course, this can confuse the consumer, so we're here today to help sort things out.
Taking advantage of that 0.22-micron process, NVIDIA has released a new variation of the TNT2, known as TNT2-A or TNT2 Pro. Internally, it is identical to the good ol’ TNT2 we’re used to, but the die shrink allows for a cooler running part with the potential for higher clock speeds. The TNT2 Pro ships at a default clock of 143/166, which slots it between the standard TNT2 and TNT2 Ultra, but closer to the Ultra.