NVIDIA GeForce GTX 690: Dual-Kepler GPUs

After teasing us for most of last week, NVIDIA has finally launched its GeForce GTX 690, a graphics card powered by two GPUs based on the “Kepler” architecture. The new card was announced at an odd hour in the USA, but that’s because Jen-Hsun Huang (NVIDIA’s CEO) presented it in Shanghai, China.

Now, you finally get to see the card in all its glory, and NVIDIA is rather proud of the reference design, which uses thixomolded magnesium alloy (instead of plastic) for the fan housing because this material provides better heat dissipation. Like previous designs, the heatsink uses vapor chambers, which is not too much to help cool the “beast”.

In ideal conditions, the GeForce GTX 690 should perform close to 2X faster than the GeForce GTX 680, but in reality, performance does not scale linearly with the number of GPUs. If you have a specific game in mind, wait for the independent benchmarks, but most likely, you will be able to crank the resolution and image quality to the maximum settings. All in all, the GeForce GTX 690 embarks 3072 CUDA cores, which are basic processing building blocks. But that’s not it, you can pair two GeForce GTX 690 to achieve a -mind blowing- total of 6144 CUDA cores.

The card should be available “by May 7”, “in limited quantities”, says NVIDIA. You can expect NVIDIA’s closest partners to have those “bragging rights” cards, namely (but not limited to) ASUS, EVGA, Gainward, Galaxy, Gigabyte, Inno3D, MSI, Palit and Zotac. Finally, the card may be the fastest in the world, but it won’t come for cheap, the price is… $999, which seems to be the price for the ultimate graphics experience of the moment.