So yes, you take a GTX 980, chop it in half in every way and that is the GeForce GTX 960. A lot of you guys will feel that this is a bridge too far. Especially the 128-bit wide bus seems to be a nag, then again Maxwell makes efficient use of memory color compression. The 960 series is based on Maxwell architecture, yes, named after the mathematical physicist. The Maxwell family of GPUs is actually the 10th generation of GPU architecture for Nvidia. With several design goals in mind (higher performance and lower power consumption) Nvidia was hoping to reach 20 nm by the time their high-end product would be released. It is now January 2015 and it is abundantly clear that the 20 nm fab nodes are a huge yield mess, as no manufacturer dares to use it. Nvidia went with plan B and stuck with a 28 nm process, this makes their silicon sizeable, in relative proportions of course. Nonetheless, Nvidia has moved forward and today the 3rd Maxwell based product (GTX 750 was actually the first trial) is being released as a GM206 based GPU. Armed with voltage, power and load limiters, Nvidia these days can harvest massive performance out of chips when you think about it. Today is about the GeForce GTX 960 range of performance. The base clock speed of the GeForce GTX 960 is 1126 MHz. The typical Boost Clock speed is 1178 MHz. The GPU used thus is still on 28 nm.
