Price: About $170

Recently, NVIDIA has begun making a new version of the TNT2 and we don't mean the GeForce 256. We're talking about the TNT2-A, a .22 micron version of the TNT2 (as opposed to the old size of .25 micron). Gigabyte Technology, a Taiwan based graphics card, motherboard, and platform solution maker has taken the TNT2-A, attached it to what appears to be an essentially stock NVIDIA reference design, and then slapped two heat sinks and a fan onto it for extreme cooling. We put the value and performance oriented GA660 Plus through the paces to see how it performs, then we overclocked it to see how far dual heat sinks and a fan could help us push the TNT2-A. Read on to see how it performed.

So what is up with the TNT2-A? The TNT2-A is essentially identical in design to the TNT2 that we all know and love, except that it uses a .22 micron fab as opposed to the original TNT2's .25 micron fab. The advantages of a .22 micron fab are many. First, .22 micron chips take up less space, so more chips can be made on the same wafer at the same time. When you can make more chips at the same time each chip costs less to make, so the cost of making a TNT2-A is lower than the cost of making a TNT2 because of the smaller .22 fab. Second, .22 micron chips have narrower and shorter circuitry, which lowers overall resistance in the chip, which means less heat. Less heat means better over clocking speeds are attainable. So the average .22 micron TNT2-A should theoretically be able to clock faster than the average .25 micron TNT2. Lower costs and faster speeds make the TNT2-A and its smaller die size a good thing.