Shootouts

NVIDIA GeForce GTX 680 SLI vs. AMD Radeon HD 7970 CrossFire

By Wong Chung Wee - 9 Apr 2012

Crysis Warhead & Far Cry 2 Results

Crysis Warhead Results

On Crysis Warhead, we noted that the AMD Radeon cards put in better performances than the NVIDIA camp, seeing that their scores consistently better than their NVIDIA counterparts regardless as a single GPU or in a 2-way CrossFire setup. From past experience, we find that AMD Radeon HD cards have always performed well in this benchmark. In this particular situation, the pair of AMD Radeon HD 7970 cards outperformed the NVIDIA pair by a margin in the range of 13% to 18% at the highest resolution while at resolution of 1920 x 1200, the margin is in the range of 7% to 13%.

In its SLI setup, the pair of Kepler GPUs scale in the range of 75% to slightly over 90% over a single GPU. Compared to the CrossFire setup that scaled in the range of 103% to 112% at the resolution of 2560 x 1600, the results here seem to point to GCN architecture as more attuned for multi-GPU setups.

And to peer in a little deeper as to how the GeForce GTX 680 manages GPU Boost, we captured the following graphs from the eVGA utiltiy. GPU1 reaches its maximum core clock speed of 1100MHz while the second Kepler GPU reaches its maximum clock rate of 1137MHz while its power consumption reached 105%.

   

 

Far Cry 2 Results

The NVIDIA GeForce GTX 680 2-way SLI setup edges out the Radeon HD 7970 CorssFire on Far Cry 2. With 4x AA enabled, the GeForce GTX 680 was about 15% faster than the Radeon HD 7970. With 8x AA enabled, the GeForce GTX 680 increases its advantage to more than 20%. In fact, on the lower resolutions of 1920 x 1200 pixels resolution, the single GeForce GTX 680 was about on par with the previous generation dual-GPU cards.

Although the pair of Kepler GPUs put in better raw scores for Far Cry 2, we observed that the pair of AMD Radeon HD 7970 cards scale better than their NVIDIA counterparts. At the highest resolution, the Southern Islands scale in the range of 63% to 93% while the Kepler GPUs continue to play catchup in their scaling performance that range from 46% to 67%.

A couple of detailed monitoring results for the pair of Keppler cards - GPU1 reaches its maximum core clock speed of 1100MHz while the second Kepler GPU reaches its maximum clock rate of 1137MHz. We note that both their power consumption rates never reached over 100% for the duration of this benchmark.

   
Join HWZ's Telegram channel here and catch all the latest tech news!
Our articles may contain affiliate links. If you buy through these links, we may earn a small commission.