Nvidia Corp. on Tuesday unveiled its second chip that belongs to Fermi generation of graphics processing units (GPUs) and also unveiled its first DirectX 11 chip for mobile computers. But while the GeForce GTX 480M is positioned as the top-of-the-range mobile offering from Nvidia, declared raw-speed performance of the solution is below that of ATI Mobility Radeon HD 5870.

“We have now brought the vaunted Fermi architecture to notebooks. GeForce GTX 480M delivers nearly three times more Nvidia CUDA cores over previous generation notebook GPUs, which means users get unbelievably fast video transcoding, upscaling from standard definition to high definition and real-time movie clean-up with the click of a button,” said Matt Wuebbling, senior product manager of notebook products.

Nvidia GeForce GTX 480M seems to be based on the code-named GF104 processor, which is a cut down version of the original GF100 first shown in September, 2009. The new chip sports 352 stream processors (SPs), 44 texture units (TUs), 32 render back ends (RBEs), 256-bit memory controller along with advanced feature-set. Thanks to decreased complexity of the GF104 processor compared to the predecessor, the new GPU can operate at 850MHz, a rather unprecedented speed for the GF100. Nvidia did not declare thermal design power of the new solution, but earlier reports suggested that the TDP of the model GTX 480M is about 100W.

Thanks to boosted frequencies, peak theoretical computing power of the GeForce GTX 480M is 897GLOPS. Still, the chip will not be the most powerful mobile GPU on the planet since ATI Mobility Radeon HD 5870 boasts with peak computing power of 1120GLOPS. Nevertheless, the model GTX 480M sports a number of capabilities its rival does not have, e.g. stereoscopic-3D support, PhysX support and so on. Yet, unlike the model GTX 480M, ATI Mobility Radeon HD 5870 consumes 50W, two times less than the rival.

The first notebooks featuring Nvidia GeForce GTX 480M will be released by Clevo, a well-known maker of high-performance notebooks.

Tags: Nvidia, Fermi, Geforce, GF100, GF104, 40nm