Nvidia’s GeForce GTX 780 utilises the same next-generation GK110 GPU architecture found on the range-topping GTX Titan. Does this equate to a Titanic performance gain?

The goal of the Kepler GK110 architecture is to deliver better processing performance and provide a more efficient graphics card that consumes less power and generates less heat. A performance boost for the successor to the GTX 680 (review) is pretty much guaranteed by the fact that the GTX 780 boasts 50% more memory and 50% more CUDA cores than its predecessor. It sports 2304 CUDA cores compared to the 1536 on offer on the GTX 680, and 3 GB of GDDR5 RAM.

Connectivity and power needs

In terms of connectivity, you get a DVI-I plus DVI-D connector; HDMI out that users with full HD monitors will appreciate; and a standard Displayport. Unfortunately the leap in performance comes with a need for more power, as the GTX 780 sucks up to 300 W via a 6-pin and an 8-pin power connector, up from the 195 W required by the GeForce GTX 680 from two 6-pin connectors.

Luckily all this translates into buttery smooth gaming performance, whether you’re playing less demanding titles such as Need for Speed: Most Wanted or graphically taxing games the likes of Crysis 3 (review). We tested the GPU on a PC powered by Intel’s Core i7-4770K CPU clocked at 3.5 GHz and that boasts 8 GB RAM. During benchmarking the GeForce GTX 780 delivered a sublime score of 3277 within Heaven Benchmark 2.0 compared to its predecessor’s tally of 2169. In addition, a score of 4259 was obtained via Futuremark’s 3DMark 2011 DX11 benchmark test, easily beating the 3151 benchmark set by the GTX 680 last year.

To the point

The performance of the Asus GeForce GTX 780 sees it punching in a higher GPU weight class than its GTX 680 predecessor. However the added performance bite does come with a bitter aftertaste - its R10 500 asking price. It’s a great deal more than the R6 220 that you fork out for last year’s GTX 680.