17seconds :
That's for the reference model GTX560 at 822 mhz. You actually would have a hard time finding one of those. Most GTX 560's are clocked around 900 mhz or more, the Gigabyte GTX560 SOC is clocked at 1000 mhz. It makes a big difference.
http://www.guru3d.com/article/gigabyte-gtx-560-ti-soc-review/
The site im getting it from has this info about the card (and its the only GTX560 ti available)
Technical data
* GPU: NVIDIA GeForce GTX 560 Ti (GF114)
* Manufacturing process: 40nm
* Maximum power consumption: 180W
* Interface: PCI Express 2.0
* Memory: 1024MB GDDR5
* GPU Clock Speed: 822MHz
* Shader Clock: 1644MHz
* Memory Clock: 4008MHz
* Memory Interface: 256 bit
* Stream Processors: 384
* Texture units: 64
* DirectX 11, Shader Model: 5.0
* Connectors: 2 Dual-Link DVI-I, mini-HDMI
* HDCP, NVIDIA SLI, PhysX, CUDA
So as I understand, that's the one used in the bench.
And for the HD6950 it has this:
Technical data
* GPU: AMD Radeon HD 6950 (Cayman Pro chip)
* Manufacturing process: 40nm
* Maximum power consumption: 200W / 20W (Load / Idle)
* Interface: PCI Express 2.0
Memory: 2048MB GDDR5
* GPU Clock Speed: 800MHz
* Memory Clock: 5000MHz
* Memory Interface: 256 bit
* Stream Processors: 1408
* Texture units: 88
* DirectX 11, Shader Model: 5
* Built-in 7.1 HD audio controller
* Connections: 1x DVI-I, 1x DVI, 1x HDMI (1.4), 2x mini-DisplayPort
* HDCP, ATI CrossFire, ATI Eyefinity
So which one is faster, now that you know the clock speeds?