Oak Ridge National Laboratory’s Titan supercomputer is currently the number 1 in the list of the Top500 supercomputers in the world. As you may know, the Titan packs 18,688 NVIDIA Tesla K20 GPU accelerators; and now NVIDIA is bringing the Titan power to the consumer market introducing the GeForce GTX TITAN GPU, which is the fastest in the world and built with the same Kepler architecture.
As The Verge pointed out, the GTX TITAN may be the world’s fastest GPU, it may not be the fastest graphics card. While the TITAN packs 2688 CUDA cores running at 846Mhz, NVIDIA’s own GeForce GTX 690 has 3072 cores at 915MHz by combining two GPUs. The TITAN gets much more video memory, 6GB of GDDR5 RAM, though. The GPU provides 4.5 Teraflops of single precision and 1.3 Teraflops of double precision processing power and supports GPU Boost 2.0 technology to automatically boost graphics performance. The card features an aluminum chassis and high efficiency vapor chamber cooling.
NVIDIA said that three GeForce GTX TITAN can be combined in 3-way SLI mode for ultimate performance to handle most demanding PC gaming titles. The GTX TITAN uses PCI Express 3.0 slot and offers one Dual-link DVI-I, one Dual-link DVI-D, a HDMI and a DisplayPort connectors with support for 4096×2160 maximum digital resolution and output to four monitor. OpenGL 4.3, 3D Vision, CUDA, DirectX 11, PhysX, TXAA, Adaptive VSync, FXAA, 3D Vision Surround are also supported.
NVIDIA’s GeForce GTX TITAN will be available on 25 February for $999 and will be available in fully configured systems from gaming PC makers.