Sign in with
Sign up | Sign in

Test Setup And Benchmarks

GeForce GTX 660 Ti Review: Nvidia's Trickle-Down Keplernomics

The closest-performing card to the GeForce GTX 660 Ti in Nvidia’s prior-generation line-up is GeForce GTX 580. So, we obviously wanted to include it in our tests. Of course, we also want to see how this new model compares to GeForce GTX 670 as well.

AMD's similarly-priced Radeon HD 7870 is an obvious contender, as is the more expensive Radeon HD 7950. We're throwing the Radeon HD 7970 in as a reference point, too. We had MSI's R7970 Lightning and R7950 on-hand to represent those two products, and we dropped each card's clock rates to reference levels. MSI supplied the GeForce GTX 680 Lightning specifically for this review, and we tested it at stock and factory overclocked settings, too.

From here, things get more complicated. AMD is attempting a preemptive strike against GeForce GTX 660 Ti in the form of Radeon HD 7950 Boost Edition. Essentially, this is a BIOS update to the vanilla Radeon HD 7950 that overclocks the board from 800 to 850 MHz and implements a 925 MHz boost state. The accelerated frequency is adjustable in the Catalyst Control Center's overdrive tool.

We’re not going to spend a lot of time talking about the Radeon HD 7950 Boost Edition because, frankly, we're not fans of such last-minute retaliations. Company representatives claim 7950 Boost Edition cards will be available by Friday. But the last time we were promised availability of a Boost Edition card (the 7970), AMD was about a month too ambitious. We'll believe this one when we see it. You'll find the card's results in our overclocking tests as a result.

Speaking of overclocking, we're including MSI’s GeForce GTX 660 Ti a second time, its core accelerated an impressive 150 MHz and memory running 250 MHz faster thanks to MSI's Afterburner tool. We were able dial in an extra 100 millivolts on the core and set 114% of the maximum board power in order to get there.

Finally, we are looking to find the highest playable settings in each game that deliver a minimum of 30 FPS and an average of at least 40. We don't want to waste GPU performance on rendering hundreds of frames per second, and we aren't interested in numbers that aren't playable.

Test System
Intel Core i7-3960X (Sandy Bridge-E), 3.3 GHz, Six Cores, LGA 2011, 15 MB Shared L3 Cache, Hyper-Threading enabled, Overclocked to 4.2 GHz
ASRock X79 Extreme9 (LGA 2011) Chipset: Intel X79 Express
On-Board Gigabit LAN controller
Corsair Vengeance LP PC3-16000, 4 x 4 GB, 1600 MT/s, CL 8-8-8-24-2T
Zotac GeForce GTX 580 AMP2 Edition
815 MHz GPU, 3 GB GDDR5 at 1025 MHz

MSI GeForce GTX 660 Ti
1019/1097 MHz Base/Boost GPU, 2 GB GDDR5 at 1502 MHz

Gigabyte GeForce GTX 670 Overclocked
980/1058 MHz Base/Boost GPU, 2 GB GDDR5 at 1502 MHz

Reference Radeon HD 7870
1000 MHz GPU, 2 GB GDDR5 at 1200 MHz

MSI Radeon R7950
800 MHz GPU, 3 GB GDDR5 at 1250 MHz

Reference Radeon HD 7950 Boost Edition
850/925 MHz Base/Boost GPU, 3 GB GDDR5 at 1250 MHz

MSI Radeon R7970 Lightning
925 MHz GPU, 3 GB GDDR5 at 1375 MHz

All overclocked cards reduced to reference specification for testing
Hard Drive
Samsung 256 GB (SSD)
ePower EP-1200E10-T2 1200 W
Software and Drivers
Operating System
Microsoft Windows 7 x64, Service Pack 1
DirectX 11
Graphics Drivers
Nvidia: 305.37 beta
AMD: Catalyst 12.7 beta
Battlefield 3
Campaign Mode, "Operation Swordfish" 60 second Fraps
Crysis 2
Adrenaline Crysis 2 Benchmark Tool, Times Square, DirectX 11
Batman: Arkham Asylum
Version, Built-in DirectX 11 Benchmark
Metro 2033Full Game, Built-In Benchmark, "Frontline" Scene
DiRT Showdown
Version, DirectX 11 Benchmark
The Elder Scrolls V: Skyrim
Update 1.4.27, THG Benchmarks save, 25 seconds Fraps
Max Payne 3
Version, Chapter 3, save 16, 65 second Fraps
React To This Article