It has 75% more CUDA cores than GTX 680 and GPU Boost 2.0Most of the relevant information on the NVIDIA GeForce GTX Titan graphics card came out yesterday, but the company has only now made the official introduction, after a one-day delay.
NVIDIA GeForce GTX Titan is, simply put, the mightiest consumer single-GPU graphics card that NVIDIA has ever made.
Even with young drivers, the overall improvement over the GeForce GTX 680 is of 30%, and NVIDIA has even published a comparison chart of GTX Titan and GTX 690, though with a catch.
Basically, NVIDIA has tested some games on a quad-SLI system using GeForce GTX 690, and a 3-way SLI system using GTX Titan.
As one might guess, the latter did better in all instances, by a lot. On average, the games ran 75% better on the Titan-enabled computer (the system used an Intel Core i7 3960X 6-core 3.3 GHz CPU and 4x MSAA at highest graphics settings).
In some games, it did 100% better (100 percent faster in Max Payne 3), while others showed less improvement (40 percent faster in TESV: Skyrim), but the point stands.
For those who want to know the numbers, GeForce GTX Titan has 2,688 CUDA cores, a base clock speed of 837 MHz, a GPU Boost clock speed of 876 MHz and floating point performance of 1.3 Teraflops (double precision) and 4.5 Teraflops (single precision).
Then, there is the memory of 6 GB GDDR5 VRAM, the interface of 384 bits and the output design: two dual-link DVI ports, one HDMI and one DisplayPort.
The high-end card works on the energy provided by two PCI Express power inputs (one 6-pin and one 8-pin) and has a TDP (thermal design power) of 250W.
The Adaptable Vsync is one of the special features on the card, but we've already explored it here, where we have also written of the GPU boost 2.0 technology.
Go here to see a benchmark between the newcomer and AMD's Radeon HD 7970, and stay tuned as we discover more of the Titan's special abilities.
NVIDIA will sell a limited number of GeForce Titan graphics cards for $1,000 / 747-1,000 Euro.