GeForce GTX 660 vs Geforce GTX 670

Intro

Compare all that to the Geforce GTX 670, which uses a 28 nm design. nVidia has clocked the core frequency at 915 MHz. The GDDR5 RAM runs at a frequency of 1500 MHz on this particular card. It features 1344 SPUs along with 112 TAUs and 32 Rasterization Operator Units.

Display Graphs

Hide Graphs

Benchmarks

These are real-world performance benchmarks that were submitted by Hardware Compare users. The scores seen here are the average of all benchmarks submitted for each respective test and hardware.

3DMark Fire Strike Graphics Score

Geforce GTX 670		7351 points
GeForce GTX 660		5063 points
	Difference: 2288 (45%)

Power Usage and Theoretical Benchmarks

Power Consumption (Max TDP)

GeForce GTX 660		140 Watts
Geforce GTX 670		170 Watts
	Difference: 30 Watts (21%)

Memory Bandwidth

In theory, the Geforce GTX 670 should be 33% quicker than the GeForce GTX 660 in general, because of its greater bandwidth. (explain)

Geforce GTX 670		192000 MB/sec
GeForce GTX 660		144192 MB/sec
	Difference: 47808 (33%)

Texel Rate

The Geforce GTX 670 is a lot (more or less 31%) better at AF than the GeForce GTX 660. (explain)

Geforce GTX 670		102480 Mtexels/sec
GeForce GTX 660		78400 Mtexels/sec
	Difference: 24080 (31%)

Pixel Rate

The Geforce GTX 670 is much (approximately 24%) more effective at full screen anti-aliasing than the GeForce GTX 660, and also able to handle higher resolutions better. (explain)

Geforce GTX 670		29280 Mpixels/sec
GeForce GTX 660		23520 Mpixels/sec
	Difference: 5760 (24%)

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.

Price Comparison

Specifications

Memory Bandwidth: Bandwidth is the max amount of information (in units of MB per second) that can be transported past the external memory interface within a second. It is worked out by multiplying the interface width by its memory clock speed. In the case of DDR memory, it must be multiplied by 2 once again. If DDR5, multiply by 4 instead. The higher the memory bandwidth, the better the card will be in general. It especially helps with anti-aliasing, High Dynamic Range and higher screen resolutions.

Texel Rate: Texel rate is the maximum texture map elements (texels) that can be processed per second. This figure is calculated by multiplying the total amount of texture units by the core clock speed of the chip. The higher this number, the better the graphics card will be at handling texture filtering (anisotropic filtering - AF). It is measured in millions of texels in one second.

Pixel Rate: Pixel rate is the maximum number of pixels the graphics card can possibly write to its local memory in one second - measured in millions of pixels per second. The figure is worked out by multiplying the number of colour ROPs by the the core clock speed. ROPs (Raster Operations Pipelines - also called Render Output Units) are responsible for outputting the pixels (image) to the screen. The actual pixel rate also depends on quite a few other factors, most notably the memory bandwidth of the card - the lower the memory bandwidth is, the lower the ability to get to the max fill rate.