GeForce GTX 650 Ti
The GeForce GTX 650 Ti is based on the same GK106 GPU as the recently tested GeForce GTX 660. Below is a block diagram of the chip. The GK106 has 960 Cuda cores divided over five SMX units, which in their turn make up three GPCs. The GK106 has a 192-bit GDDR5 memory controller, which is a combination of three 64-bit segments.
Nvidia disabled certain features on the GPU in the GTX 650 Ti. Out of the five SMX units, Nvida deactivated one. That leaves 768 shader units or Cuda cores and 64 texture units. That's a reduction of 20 percent compared to the GTX 660, but twice that of the non-Ti GeForce 650. The memory controller also has one of its three 64-bit segments deactivated, resulting in a 128-bit controller. This is pretty standard for graphics cards in this segment.
Nvidia has set the standard clock frequency for the GPU at 925 MHz, a bit less than the 980 MHz of the GTX 660. Another important aspect is that the GTX 650 Ti has a turbo mode and therefore no Boost clock. The memory runs at a standard frequency of 1350 MHz.
The card has a TDP of 110 watts, according to Nvidia, and only needs a single 6-pin PEG connector. A PC with a 400 watt PSU should be sufficient here. The reference card has two dual-link DVI ports and one HDMI connector. You don't get any SLI connectors on the GTX 650 Ti.
The remaining features of the GK106 are identical to those of the GK108 chip, including TXAA, Adaptive V-Sync, 3D Vision Surround en NVENC. We won't cover these again, please read our GeForce GTX 680 review for in-depth descriptions.
The chart below lists the differences between the GeForce GTX 650 Ti, GTX 660 and GTX 660 Ti.