GEFORCE GTX 650 Ti
Although the GeForce GTX 660 has a lower price and power compared to top-end video cards, there is still too much of a gap between them and budget GK107-based boards, which is being filled by the recently announced GeForce GTX 650 Ti video card. The suffix given to it means that this board is faster than the regular GTX 650 based on the GK107, but slower than the GTX 660 we previously reviewed, based on the GK106 chip.
It is quite logical that the new GeForce GTX 650 Ti model in NVIDIA's product line is located between the GeForce GTX 660 and GTX 650. It replaces the previous generation of mid-range boards, like the GeForce GTX 560 and GTX 550 Ti, offered before the release of solutions based on the Kepler architecture. The principle of naming NVIDIA video cards has not changed, the budget model received the number “5” in the middle of the index, but since this is the fastest of the two budget GTX 650, the suffix “Ti” was also attached to the name, which distinguishes the new product from the GTX 650 based on GK107, which we will talk about later.
The release of a video card based on a cut-down GK106 chip indicates that the production has already accumulated some defective chips from chips that did not pass the GTX 660, but which can be used for the GTX 650 Ti. As for the solutions of the competing company AMD, the new GTX 650 Ti has no direct competitor. The recommended price of the new product fits between the prices of the Radeon HD 7770 and HD 7850 models, but is still much closer to the former.
GeForce GTX 650 Ti specifications
Name | GeForce GTX 650 Ti |
Core | GK106 |
Process technology (µm) | 0.028 |
Transistors (millions) | 2540 |
Core frequency | 925 |
Memory operating frequency (DDR) | 2700 (5400) |
Bus and memory type | 128-bit GDDR5 |
Bandwidth (Gb/s) | 86 |
Unified shader units | 768 |
Unified shader unit frequency | 925 |
TMU on conveyor | 64 |
ROP | 16 |
DirectX | 11.1 |
Memory Capacity | 1024 |
Interface | PCI-E 3.0 |
So, in the case of the GeForce GTX 650 Ti model, four of the five physically available SMX units are active in the GK106 chip, which in total amounts to 768 stream processors and 64 texture units. Unlike the GeForce GTX 660, which has all three 64-bit memory controllers active, the GeForce GTX 650 Ti memory subsystem uses only two 64-bit controllers out of the three available on the chip, which in total amounts to only a 128-bit bus.
And the video memory frequency was reduced from 6000 to 5400 MHz. As a result, the resulting throughput of 86,4 GB/s may not be enough, given the power of the GK106 and the high bandwidth requirements of modern games, especially with full-screen anti-aliasing enabled. Since ROP raster operation blocks are “tied” to memory controllers, their number has also changed compared to the GTX 660, and in our case it is 16 active blocks.
GeForce GTX 650 Ti does not support the technology GPU Boost. Let us remind you that this is a combination of software and hardware solutions that allows you to automatically increase operating frequencies. GPU, depending on its power consumption to achieve maximum performance. This technology ensures operation at the highest possible frequencies within the selected thermal package and works independently of drivers or other software. Apparently, NVIDIA decided that there is no need to spoil "budget" users with such complex technologies, and should be happy already at the sight of Kepler for $ 149. In general, the nominal (and only) clock frequency of the GK106 video chip used in the GeForce GTX 650 Ti is 925 MHz, and it is fixed, unlike more powerful solutions.