- Core clock 785MHz
- GT200 core
- 216 Stream Processors
- 55 nanometer process
- 1.4 Billion Transistors
- 384-bit memory interface
- DirectX 10
- Pixel Shader 4.0
- Vertex Shader 4.0
- Unified Shader
- Geometry Shader
- CUDA
- PhysX software support
- PureVideo HD
- 666MHz core clock
- 2.3GHz memory clock speed
- 896MB GDDR3 memory
- 448-bit memory interface
|
Brand Name |
XFX
|
|
Part Number |
GX260NADBF
|
|
Graphics Chip |
GT200
|
|
Core clock |
666
|
|
Shader Clock |
1440
|
|
SPs |
216
|
|
Fabrication Process |
65 nanometer
|
|
Transistors |
1.4 Billion
|
|
Memory clock |
2.3GHz
|
|
Memory Interface |
448-bit
|
|
Memory bandwidth |
128.8GB/second
|
|
Memory Size |
896MB
|
|
ROPs |
32
|
|
Texture Filtering Units |
80
|
|
Texture Filtering Rate |
69.4Gigatexels/second
|
|
HDCP Support |
Yes
|
|
HDMI Support |
Yes (via adapter)
|
|
Connectors |
2x Dual-Link DVI, TV-Out
|
|
RAMDACs |
400MHz
|
|
Bus |
PCI Express 2.0
|
|
Form Factor |
Dual Slot
|
|
Power Connectors |
2x6-pin
|
The NVIDIA GeForce GTX 260 216 core architecture should look familiar to anyone having seen their launches the last few months as it is identical to the GeForce GTX260 in virtually every respect except for one, the number of Shader Processors. The original GeForce GTX 260 had 192 SPs or 6 clusters of 32 SPs each. The new GeForce GTX 260 216 has 216 SPs as the name suggests, meaning that the 7th cluster was activated for this chip out of the 8 on the GeForce GTX 280 has.
NVIDIA states that the entirety of their production of GeForce GTX 260s should be based upon the 216 core variant, meaning that the older variant will be phased out with remaining inventory sold. The GeForce GTX 260 is based upon NVIDIA's GT200 chip and has 1.4 Billion transistors on the 55 nanometer process. The GTX 260 216 was launched a short month or two ago with two companies making the initial cards. Today the other manufacturers are joining in and the card being used for this review is the XFX GeForce GTX 260 Black Edition.
The memory bandwidth on the GeForce GTX is found by multiplying the memory clock speed by the bus, multiplying that result by 2 and dividing the final result by eight. On the XFX GTX 260 Black Edition the memory bandwidth is 128.8GB/second or 448-bit/8x2.3GHzx2. This is nearly the same bandwidth that is found on the GeForce GTX 280 reference clocked cards and is higher than the GTX 260s on the market today.
NVIDIA was the first graphics card manufacturer to fully support DirectX 10.0 which was a new API in 2006 with the release of the GeForce 8800 GTX. This card supported Pixel Shader 4.0, Vertex Shader 4.0, and Geometry Shaders and was NVIDIA's first card to have a Unified Shader Architecture. Today virtually all games being released support some DirectX 10.0 features. One ulterior motive for doing this review was moving to all new benchmarks with mostly DirectX 10.0 support.
One of the more compelling features of the NVIDIA cards is the ability to support CUDA and PhysX. Using the graphics card to solve computational problems makes sense as the graphics chips are capable of TeraFLOPS computing power in the form of the HD 4850 or HD 4850 cards and the GeForce GTX 260 is nearly capable of that feat. PhysX allows the game to use the graphics card to do the physics calculations and overall performance can be done due to this. Early next year EA will be announcing the first PhysX AAA title Mirror's Edge. XFX has taken the base GeForce GTX 260 card and improved it immeasurably by increasing the clock speed from 575MHz to a whopping 666MHz. Is the GTX 260 Black Edition the Devil's card? Nope, just the fastest darn GeForce GTX 260 216 on the market today. The Shader Clock is set at a modest increase to 1440MHz, but this shouldn't be a bottleneck for this video card anyway.
