Nvidia Quadro P6000 houses full-scale GP102 with 3,840 CUDA cores

That Nvidia is working on a new worsting card under the slogan Titan did not come as a big surprise, but that the launch takes place already on August 2 and that it is not about the graphics circuit GP100 with HBM2 memory surprised. Instead, it is about GP102, which now also makes its entrance in the new calculation card Quadro P6000.

Specifications: Nvidia Quadro P6000

When the announcement of the Nvidia Titan X began, many began to suspect that it was not a full-scale version of the GP102, something that is now confirmed when Nvidia unveils the new computing card Quadro P6000. Instead of 3,584 CUDA cores, this one comes with no less than 3,840 pieces – 50 percent more than the GP104 that houses the Geforce GTX 1080.

What is also clear is that the Quadro P6000 is more powerful than the Titan X, at least on paper. Nvidia is talking about 12,000 GFLOPS, which with the number of CUDA cores in mind corresponds to a clock frequency of around 1,560 MHz. Despite this, the card’s stated power consumption is the same at 250 W, something that can possibly be explained by the fact that the memory frequency dropped back to 9,000 from 10,000 MHz for Titan X.

The fact that the Nvidia Titan X will not be equipped with a full-scale GP102 probably means that the best circuits are hand-picked for the Quadro P6000, but also that a more powerful graphics card for consumers may be in the plans later on. Such a strategy would be reminiscent of the launch of the original Geforce GTX Titan, which was released with a scaled-down version of the GK110 graphics circuit, to later replace it with the GTX Titan Black with a full-scale variant.

Alongside the Quadro P6000, the P5000 is also announced, which is roughly a Geforce GTX 1080 with 16 GB of GDDR5X at 9,000 MHz. The launch of the duo takes place in October and what the prices will land on is not yet known.