Although not apparent in the above diagram, the architectural efficiency of the
GeForce GTX 200 GPUs is substantially enhanced over the prior generation. We’ll
be discussing many areas that were improved in more detail, such as texture
processing, geometry shading, dual issue, and stream out. In directed tests, GeForce
GTX 200 GPUs can attain efficiencies closer to the theoretical performance limits
than could prior generations.
Table 2 compares the GeForce 8800 GTX to the new GeForce GTX 280 GPU.
You will notice sizable increases in a number of important measurable parameters.
Features
8800 GTX
GTX 280
% Increase
Cores
128 240 87.5
%
TEX
64t/clk 80t/clk 25
%
ROP Blend
12p/clk 32p/clk 167
%
Precision
fp32 fp64 --
GFLOPs
518 933 80
%
FB Bandwidth
86 GB
142 GB
65 %
Texture Fill
37 GT/s
48 GT/s
29.7 %
ROP Blend
7 GBL/s
19 GBL/s
171 %
PCI Express
6.4 GB
12.8 GB
100 %
Video
VP1 VP2 --
Table 2: GeForce 8800 GTX vs GeForce GTX 280
May 2008 | TB-04044-001_v01
11
