NVIDIA has finally come forth and released its latest video board, a high-end graphics card known as GeForce GTX 580 and powered by the Fermi architecture, the company's GPU technology that supports DirectX 11 graphics.The GeForce GTX 580 is based on the GF110 graphics processing unit (GPU), the Santa Clara, California-based company's latest Fermi chip.
As such, it has 512 CUDA cores, their frequency being of 1,544 MHz, while the GPU itself operates at a clock speed of 772 MHz.
Additionally, the newcomer features 1,536 MB of GDDR5 VRAM, whose own frequency is of 4.0 Gbps (4,008 MHz to be exact). Furthermore, said amount of high-grade memory has an interface of 384 bits.
As far as design goes, the GeForce GTX 580 has a length of 10.5 inches and a dual-slot fansink, with adaptive GPU fan control and a custom vapor chamber.
The cover design is even optimized for SLI, so that multi-GPU configurations (3-way SLI is available) may be more easily set up.
When gaming, the newcomer is supposed to provide faster FPS and high geometric realism even while the current and voltage on each 12V rail is monitored in real time (there are one 8-pin and one 6-pin PCI Express connectors).
What's more, during highly demanding applications like Futuremark and OCCT, the performance is dynamically adjusted so as to keep power within specs (power consumption is of 244W).
Other enhancements, compared to the GTX 480, include extra function units for tessellation, shading and texturing.
NVIDIA even states that the adapter also integrates architectural enhancements for full speed FP16 texture filtering.
Needless to say, CUDA and PhysX are supported, as is the 3D Vision technology, while video output options include HDMI and dual-DVI.
All in all, the GeForce GTX 580 is a faster, more efficient and more silent successor to the GTX 480 and should start selling immediately, for $499 in the US and 479 Euro in Europe.