Compare any two graphics cards:
GeForce 8800 GTS (G80) 320MB vs GeForce GTS 450
IntroThe GeForce 8800 GTS (G80) 320MB has core speeds of 513 MHz on the GPU, and 792 MHz on the 320 MB of GDDR3 memory. It features 96 SPUs as well as 48 TAUs and 20 Rasterization Operator Units.
Compare all that to the GeForce GTS 450, which features core speeds of 783 MHz on the GPU, and 902 MHz on the 512 MB of GDDR5 memory. It features 192 SPUs along with 32 Texture Address Units and 16 ROPs.
Power Usage and Theoretical Benchmarks
Power Consumption (Max TDP)
The GeForce 8800 GTS (G80) 320MB should in theory be just a bit faster than the GeForce GTS 450 overall. (explain)
Texel RateThe GeForce GTS 450 should be just a bit (more or less 2%) faster with regards to AF than the GeForce 8800 GTS (G80) 320MB. (explain)
Pixel RateIf running with high levels of AA is important to you, then the GeForce GTS 450 is the winner, by a large margin. (explain)
Please note that the above 'benchmarks' are all just theoretical - the results were calculated based on the card's specifications, and real-world performance may (and probably will) vary at least a bit.
Memory Bandwidth: Bandwidth is the max amount of data (counted in megabytes per second) that can be moved past the external memory interface in a second. It's calculated by multiplying the bus width by the speed of its memory. If the card has DDR RAM, the result should be multiplied by 2 again. If DDR5, multiply by 4 instead. The higher the memory bandwidth, the better the card will be in general. It especially helps with AA, HDR and higher screen resolutions.
Texel Rate: Texel rate is the maximum number of texture map elements (texels) that are processed in one second. This number is worked out by multiplying the total amount of texture units by the core speed of the chip. The higher the texel rate, the better the card will be at texture filtering (anisotropic filtering - AF). It is measured in millions of texels per second.
Pixel Rate: Pixel rate is the maximum number of pixels the graphics card can possibly write to its local memory per second - measured in millions of pixels per second. The figure is calculated by multiplying the number of Raster Operations Pipelines by the clock speed of the card. ROPs (Raster Operations Pipelines - aka Render Output Units) are responsible for drawing the pixels (image) on the screen. The actual pixel output rate is also dependant on many other factors, most notably the memory bandwidth - the lower the bandwidth is, the lower the potential to get to the maximum fill rate.