|
|
In English
SuomeksiCompared
Just a quick comparison on the physical dimensions. The 9800GX2 and the GTX 280 are more or less equally long, around 27 centimeters. That is around four centimeters longer than for example the 8800 GTS. Power connectors are located on the side so that the cards will be usable on the cases that are not that deep.
GTX280 in numbers
| GeForce 8800 GTX |
GeForce 8800 GT |
GeForce 8800 GTS (G92) |
GeForce 9800GX2 |
GeForce GTX260 |
GeForce GTX280 |
|
| Stream Processors | 128 | 112 | 128 | 256 | 192 | 240 |
| Core Clock (MHz) | 575 | 600 | 678 | 600 | 576 | 602 |
| Shader Clock (MHz) | 1350 | 1500 | 1700 | 1500 | 1242 | 1296 |
| Memory Clock (MHz) | 900 | 900 | 986 | 1000 | 999 | 1107 |
| Memory Amount | 768MB | 512 MB | 512 MB | 1024 MB | 896 MB | 1024 MB |
| Memory Interface | 384-bit | 256-bit | 256-bit | 512-bit (2*256) | 448-bit | 512-bit |
| Memory Bandwidth (GB/sec) | 86.4 | 57.6 | 63.1 | 128.0 | 111.9 | 141.7 |
| Texture Fill Rate (billion/sec) | 36.8 | 33.6 | 38.0 | 76.8 | 36.9 | 48.2 |
What immediately pops up is the true 512-bit memory interface, using eight 64-bit wide frame buffer interface units. Only thing we need to know is that the memory bandwidth has been significantly increased.
When talking about numbers, one has to note this important change on the GTX 200 architecture. Here is what NVIDIA has to say about their move to the 64-bit double-precision:
Double Precision Support
A very important new addition to the GeForce GTX 200 GPU architecture is
double-precision, 64-bit floating point computation support. This benefits various
high-end scientific, engineering, and financial computing applications or any
computational task requiring very high accuracy of results. Each SM incorporates a
double-precision 64-bit floating math unit, for a total of 30 double-precision 64-bit
processing cores.
The double-precision unit performs a fused MAD, which is a high-precision
implementation of a MAD instruction that is also fully IEEE 754R floating-point
specification compliant. The overall double-precision performance of all 10 TPCs of
a GeForce GTX 200 GPU is roughly equivalent to an eight-core Xeon CPU,
yielding up to 90 gigaflops.
While we move forward on some areas, it is bit odd to learn that we took a step back...
Note: DirectX 10.1 Not Supported
DirectX 10.1 is not supported in GeForce GTX 200 GPUs. DirectX 10.1 includes incremental
feature additions beyond DirectX 10, some of which GeForce 8/9/200 GPUs already support
(multisample readback for example). We considered DirectX 10.1 support during the initial GPU
design phase and consulted with key software development partners. Feedback indicated DirectX
10.1 was not important, so we chose to focus on delivering better performance and architectural
efficiency.
If one hasn't followed the physical growth of the GPUs lately, it would be easy to miss the enormous size of this
chip. The 65nm GPU that is made by the Taiwan Semiconductor Manufacturing Company (TSMC) is their largest
chip to the date. 1.4 billions transistors do need their space.
 | Pages: 1 2 3 4 5 6 7 |  |
Content in english!
Sisältö suomeksi!
