NVIDIA's Surprise is a Dual-GPU GeForce GTX 690
A while ago, NVIDIA posted a simple image with two words on their Facebook page: “It’s Coming.” Rumors floated around about either a GTX 690 or a mid-range card with the Kepler GK104 architecture. Well, it was announced over the weekend their big surprise: the GTX 690. It’s a dual-GPU videocard with close-to-identical performance with two GTX 680’s in SLI, but with much less noise and heat.
So what exactly is in this thing? It’s basically two GTX 680’s on a single board, bringing a total of 3072 CUDA cores to the field. The full specifications are below:
| CUDA Cores | 3072 |
| Base Clock | 915 MHz |
| Boost Clock | 1019 MHz |
| Memory Configuration | 4GB / 512-bit GDDR5 |
| Memory Speed | 6.0 Gbps |
| Power Connectors | 8-pin + 8-pin |
| TDP | 300W |
| Outputs | 3x DL-DVI Mini-DisplayPort 1.2 |
| Bus Interface | PCI Express 3.0 |
Aside from some ultra-powerful hardware, it does some other cool things that work better on the GTX 690 than they do on the GTX 680. Those things include NVIDIA FXAA, a special kind of anti-aliasing that reduces even more jaggedness; and adaptive V-sync, which turns V-sync on and off depending on your framerate. The GTX 690 also brings improved framerate metering, which helps reduce stuttering.
Performance wise, it’s pretty astounding. While it doesn’t double the performance of the GTX 680, it gets quite close depending on the test. To read more about the card, and see some framerate benchmarks as well as more pictures, head on over to the announcement page.