NVIDIA GeForce GTX 680 brings turbo technology and comes in March
Nvidia is apparently planning to introduce in March, the first graphics card with Kepler graphics chip: The GK104-GPU GeForce GTX 680 from the 28-nanometer manufacturing binds to 2 GB of GDDR5 memory with 256 data lines. This was confirmed by several manufacturers at CeBIT told heise online.] Later versions should therefore come with 4 GB of memory.[/B The exact number of shader cores is still unclear, but it is currently out of 1536 pieces.
A significant departure from the current generation of Nvidia GPUs is the so-called dynamic clocking, or Dynamic Clock Adjustment. This turbo circuit for the graphics chip works much like Intel's Turbo Boost and Intel Turbo Core: Nvidia sets for Kepler graphics card has a maximum power consumption, they do not exceed. In games, the average power consumption is usually well below this value. Then, Kepler graphics cards increase the GPU clock speed on the fly and automatically, until they hit either of the defined limits of power consumption or to a graphics card in the BIOS stored maximum value for the 3D clock frequency. The values ​​for 3DMax and 3D are available for specifying a source in a fixed relationship to each other, a common room would be between 5 and 7 percent. Dynamic Turbo Automatic expected benchmark comparisons difficult with other graphics cards. However, it was to hear from some manufacturers, there will be the launch of the GTX 680 in late March first no-OC versions.
the term is associated with launch of the GeForce GTX 680 to see relatively more of my of H asked, "board partners", that graphics card manufacturers: They expect that Nvidia can initially deliver only small quantities of Kepler graphics chips. Some partners will initially gotten anything. Others criticized the restrictive information policy of Nvidia: "We know almost nothing, although we will sell the cards soon." Even prices on the manufacturer claims to have not been informed. This is a demonstration of the GeForce GTX 680 very shortly before: Tomorrow Nvidia wants to San Francisco a few selected journalists reveal the complete technical details. The participants of the event, just before the CeBIT invited surprising. Nvidia is currently in the defense against AMD, the competitor has already several versions of its 28nm GPUs series Radeon HD made ​​7000 with PCI Express 3.0 interface on the market, not only to DirectX 11.1 compatible and very fast , but also economically idle. The market share from Nvidia is finally dropped .
little concrete we could bring to the show on the expected performance of the 3D GeForce GTX 680 in experience. The circulating on the Internet rumors that the GTX 680, the latest AMD flagship Radeon HD 7970 by up to 25 per cent wanted to overtake, but none confirmed. Instead, it was claimed the 28-nm debut of Nvidia is Battlefield 3 in about 10 percent faster than inferior to a Radeon HD 7970 and the DirectX 11 benchmark 3DMark 11 even.
In April, followed by the GeForce GTX 680 mid-range cards with GK107 graphics chips. On the occasion of the 14th to 17 May at place GPU Technology Conference, Nvidia is planning to introduce a dual-GPU graphics card with two graphics chips GK104. GK110 from large chip that is rumored to contain 2304 shader cores, by summer is nothing to see. He should first come to Tesla supercomputer for computer cards and other HPC systems are used. ( mfi )
