NVIDIA launched its first Fermi-based GPU earlier this year in the form of the GeForce GTX 480, and it was met with mixed reception. Until now, its been the fastest single-GPU offering on the market, but certain downsides kept it from being the first-choice of many. Does NVIDIAs first proper follow-up fix all that was wrong?

SNIP: For many, it felt like Nvidia took forever to launch their GeForce GTX 480 Fermi cards, with 3.2 Billion transistors and 512 CUDA cores (480 exposed). And while the launch was great and the performance was excellent, many expected a bit more from the 480. Several early buyers were quite dissapointed with the 480s power consumption, temperatures, and noise. 7 months after the GeForce GTX 480 was launched, Nvidia is releasing their new flagship GPU, and the successor to the GTX 480: the GeForce GTX 580.

The GTX 580 does not use an entirely new GPU, but rather an optimized version of the GF100 core dubbed the GF110. This GPU consists of 4 GPCs, 16 Polymorph engines, 64 Texture units and 512 CUDA cores. Most of these specifications are similar to the GTX 480, as both have the same amount of GPCs, Polymorph engines and Texture unites, but with the GTX 580 NVIDIA has maxed out the available CUDA cores. This means through optimization, we are finally able to see the full power of the Fermi architecture.

*Quote:* *Rumors aplenty have been swirling about AMDs new HD 5900 series but NVIDIA hasnt been sitting on their laurels either. In order to preempt the release of AMDs upcoming flagship products, a new GeForce branded card is being introduced today. Dubbed the GTX 580, it uses a newly tweaked GF110 core along with higher clock speeds than the GTX 480 and the result is a class-leading card that dominates the single GPU landscape.* *Image URL:* http://images.hardwarecanucks.com/image//skymtl/GPU/GTX-580/GTX-580-1.jpg We would appreciate it if you would post this in your News section.