Testing Methodology

We chose to run the test on our Intel Core i5 2500K running at 4GHz. We decided to choose the i5 2500L for this test simply because of its immense popularity in the performance segment. The 2500K has great out-of-the-box overclocking potential and is priced sweetly. There’s a high probability that most new GTX 660 Ti owners will be in the 2500K segment, or a processor one notch up or down. This will give us results most suitable for our readers’ specifications.

The operating system we’ve used is Windows 8 Professional 64-bit. There’s not much said about Windows 8 gaming performance, minus the “App Store” gibberish, we will analyze that and see whether an upgrade from Windows 7 is really worth it, if you’re solely looking at gaming.

Benchmarks Used
Heaven Benchmark v3.0
3DMark Vantage
Battlefield 3
Total War: Shogun 2
The Elder Scrolls V: Skyrim
Batman: Arkham City

Unigine Heaven v3.

The Heaven v3.0 benchmark is built around the Unigine game engine, a cross-platform real-time 3D engine. The DX11 mode stresses tessellation, or advanced SSAO. It features volumetric cumulonimbus clouds generated by an accurate algorithm and dynamic light and sky scattering.

The Gigabyte GTX 660 Ti has no problem handling the Heaven benchmark, as is clearly evident from the numbers. The Gigabyte variant is clearly ahead of its other counterparts, though it is still beaten by the MSI GTX 660 Ti Power Edition. EVGA performs poorly, being the only card behind the Radeon HD 7950.

The single extra ROP for the GTX 670’s performance matters, as we see now. The 670 is a good 10 points ahead of the 660 Ti in average. What is surprising, though, is the score of the Radeon HD 7950. The extra memory bandwidth is clearly not scaling with real-time gaming performance. We suspect it has something to do with driver issues.