News Headlines
- Wed, Aug 17
- Square Enix confirms NieR: Automata due out on Steam in early 2017
- For Honor collector's edition announced, surprise your loved one with helmets
- Pokémon GO faces class action lawsuit over Pokémon spawns on private property
- Nioh beta begins August 23rd, promising more weapons and challenges
- Tue, Aug 16
- Twitch buys VoIP and game mod host Curse for undisclosed amount
New Articles
Related Articles

AMD has dominated the news cycles for the last few weeks thanks to the release of their latest graphics cards built on the Northern Islands architecture. The new GPUs from the red team were positioned in the mid-range market and performed very well for their price segment. What's more, in just a few weeks AMD will unveil their high-end Cayman series GPU, thus taking further mindshare. However, NVIDIA has not just been sitting idly by and allowing AMD to steal all the buzz. In fact, just a few weeks ago, rumors started to surface of a mysterious graphics card from NVIDIA which was stated to leverage the full potential of the Fermi architecture, and to be dubbed the GTX 580. All news on the matter was pure speculation, and many were not sure if this card was fact or fiction.
Today we know the GTX 580 is indeed very real, and it is designed for one purpose: to be the fastest single-GPU graphics card in the world. This is not an unfeasible accomplishment, as the GTX 480 currently holds that title, which sets the GTX 580 up to become NVIDIA's current flagship product. In addition to raw gaming power, the GTX 580 is also designed to supply better performance per watt than the GTX 480 and features a newly designed cooling solution to increase overall efficiency. Both of these are areas where the GTX 480 was hardest hit by the press, so we will be looking closely at these two aspects of the new card in the review.
Like recent products released from NVIDIA, the GTX 580 uses the Fermi architecture and as such uses a large amount of CUDA cores, a high memory bus and fast GDDR5 memory. Since it is still part of the Fermi family, it doesn't use a radically new GPU, but rather NVIDIA has improved upon what was currently available by unlocking all 512 available CUDA cores. Additionally, Nvidia has also increased the clock frequencies over the GTX 480 and improved the chip down to the transistor level. This will also enhance Tessellation and geometry performance, as well as improve the power efficiency of the GTX 580. In all, NVIDIA states that we should be seeing performance gains of 20% to 30% over the GTX 480, and increased performance per watt ranging from 25% to 35%.
With all the gaming power available to the GTX 580, it will be interesting to see exactly how well a fully equipped Fermi chip fares against the currently fastest dual GPU card available, the HD 5970.

| Specifications | |
|
Cuda Cores |
512 |
|
Gfx/Processor Clock |
772/1544MHz |
|
Memory Config |
1.5GB/ 384-bit GDDR5 |
|
Memory Speed |
4.0Gbps |
|
Power Connectors |
6-pin 8-pin |
|
Power |
244W |
|
SLI |
3-way |
|
Length |
10.5 Inches |
|
Thermal |
Dual-Slot Fansink |
|
Outputs |
DL-DVI, DL-DVI, Mini-HDMI |
Article Index |
|
Great review, colleague.
http://benchmarkreviews.com/index.php?option=com_content&task=view&id=614&Itemid=72
oh, and a side note- hdmi 1.4 is not required for 3d. high speed 1.3 has the necessary bandwidth for 3d and performs just as well for 3d as 1.4. its kind of a scam how they trick people into thinking 1.4 is needed for 3d.
Also about the 3D. I didn't say that older versions couldn't support 3D. Just that this card uses the 1.4a connection and that it does support 3D.