NVIDIA Launches Five Fermi-based Quadro Models

When NVIDIA first launched its Fermi architecture earlier this year, the crowd that seemed to gain the most consist of those who take advantage of GPU’s for non-gaming purposes, such as scientists, designers or anyone that needs to execute complex mathematical code at a heavily paralleled level. Of course, we knew it’d be only a matter of time before Fermi-based Quadro workstation cards would arrive, and they’re now here.

Taking advantage of the currently-running SIGGRAPH 2010 event, NVIDIA has launched five different Quadro models based on Fermi, which includes one mobile model that will become available this fall (5000M). The desktop variants include the Quadro 4000, 5000, 6000 and the ultra high-end 7000. Each of these cards include dual DisplayPort ports, and a single DVI, although the 7000 amps things up to offer just four DVI.

So far, we’re not going to see any entry-level Quadro cards, and I wouldn’t expect that to happen for a while. On the “low-end”, there’s the $1,199 Quadro 4000, which includes 2GB of GDDR5 and 256 cores. Next up is the $2,249 Quadro 5000, which bumps up the CUDA cores to 352 and also the memory to 2.5GB. For those really wanting to part with some heavy cash, there’s the Quadro 6000, at $4,999. This card includes a staggering 6GB of GDDR5 and 448 CUDA cores.

Then, there’s the big boy… the Quadro card that redefines “ultra high-end”. The Quadro 7000 looks to be essentially 2x Quadro 6000 put into the same graphics card, and as a result, it features 896 CUDA cores (almost double the GTX 480), and an unbelievable 12GB of GDDR5. The cost? $14,500. Yes, $14,500. It’s just unfortunate that NVIDIA hasn’t posted a picture of this card yet. It can be assumed that as these are of an extreme niche sort, they will be the last cards to become officially available.

You can read a lot more about the latest Quadro cards here, and check out the press release below.

There are significant challenges to making things ever smaller, which I am not going to address here, but here are a couple of links for further reading: Producing Integrated Circuits With X-ray Lithography and Wikipedia on Electron Beam Lithography. The bottom line is that there are limits to how small things can get with current technology. Flash densities are going to have data density growth problems, just as other storage technologies have had over the last 30 years. This should surprise no one.