A similar layout to the 5-series’ GF114 GPUs means four graphics processor clusters (GPCs) of two streaming multi-processors (SM) each, but in a predictable reduction in comparison to the GTX 680 2GB, Nvidia has disabled a single SM on the GTX 670 2GB. You can read more about the Kepler architecture in our GTX 680 2GB Review, but the GTX 670 2GB uses the same Kepler GK104 architecture of both the GTX 680 4GB and GTX 690 4GB, with the move to 28nm seeing a dramatic jump in stream processor count, but also the unification of the core and shader clock. Let’s start by taking a look at the GTX 670 2GB’s GPU. Sure, it’s still a high price to pay for a GPU, but as demand for GTX 680 2GBs is keeping prices firmly above the £430 mark, it’s still a sizable difference. It also introduces an automatic dynamic overclocking method which gives the card the leading edge, making it the fastest single GPU graphics card in the world. The card comes with massively improved performance per Watt. Today sees Nvidia attempt to chip away at this, with the GTX 670 2GB expanding the range of Kepler GPUs and bringing the fight to the £330 price point. Today NVIDIA released their new GeForce GTX 680 which is based on the companys brand-new Kepler architecture. Where AMD still holds sway though is at the more lucrative lower price points, having already launched its full product stack of 28nm parts right the way down to sub £100 prices. The GTX 670 2GB is £100 cheaper than the GTX 680 2GB
0 Comments
Leave a Reply. |