Review EVGA NVIDIA GeForce GTX 680 Kepler Review

Today we will be taking a look at the new GeForce GTX 680 graphics card, NVIDIA's latest flagship. The model we have is a reference card from EVGA, which we will be comparing to AMD's top GPU the Radeon HD 7970 and their previous offering the HD 6970. In this review we will also be performing the tests on AMD's current top end CPU and motherboard platform, which you can find more details of later on in the review.

With the arrival of the GeForce GTX 680 comes a new architecture, codenamed Kepler, which replaces the Fermi architecture of NVIDIA's two previous generations of cards, the 400 and 500 series. You can see the specifications of the new card below, alongside the other two cards we will be comparing with plus NVIDIA's previous offering the GTX 580:



Here is a screenshot of GPU-Z showing the properties of the card:



One new technology that the GTX 680 brings is called GPU Boost, this works much like the Turbo boost features on the latest Intel and AMD CPUs. A base clock is set, in the GTX 680's case this is 1006MHz, but most cards are capable of exceeding this in reasonable conditions and so the card will automatically overclock itself if it has enough headroom in terms of temperature and power draw. You can see this in the GPU-Z screenshot below, note the top line shows that during the render test, the card (and drivers) have automatically overclocked the card from the base clock of 1006 MHz, up to 1123.5 MHz. NVIDIA specify that there will be an average boost clock of 1058 MHz, to achieve this the vcore is also increased, which will increase heat output, so the card may clock back down again if it runs out of TDP (thermal design power) headroom.



On the next page of our review we will take a look at EVGA's product packaging!