The GK104 core packs 1536 CUDA cores working at 1006MHz, 1058MHz in max boost mode
The specifications of Nvidia’s upcoming GeForce GTX 680 graphics card based on the high-performance Kepler architecture were allegedly leaked a few hours ago, when a GK104 presentation slide was published online by a Chinese website.The slide, which is labeled “Nvidia Confidential,” talks about some of the basic specifications of the new GeForce video card, such as its GPU and memory frequencies or its TDP.
According to the info published by PC Online, the GK104 core, which will stand at the heart of the GTX 680, includes 1536 CUDA cores that are linked via a 256-bit wide bus to 2GB of GDDR5 memory.
Nvidia has clocked this VRAM at an impressive 1,500MHz (6GHz data rate), while the base GPU frequency is set at 1,006MHz.
As we previously revealed , Nvidia’s next-generation Kepler graphics core will also feature support for a new technology called Dynamic Clock Adjustment.
Apparently, this works similarly to Intel’s Turbo Boost technology to automatically increase the graphics core frequency to a maximum of 1058MHz, when the video card works below its rated TDP.
Speaking of TDP, the GTX 680 is expected to fit inside a 195 Watt thermal envelope and gets its power from two 6-pin PCI Express connectors that are stacked one on top of each other at the rear of the PCB.
Just like its direct counterpart, the Radeon HD 7970, Nvidia’s video card is PCI Express 3.0 compatible.
As far as the video output configuration is concerned, the GTX 680 will come equipped with a pair of dual-link DVI ports, an HDMI port, as well as with a DisplayPort 1.2 connector.
Nvidia’s first GK104-based graphics cards are expected to arrive in late March (probably March 23). Initially, graphics card makers expect Nvidia to ship only limited quantities of GK104 chips, which means that widespread availability isn’t expected before Q2 2012.