Workstations will now be able to use Nvidia’s Quadro GP100, which will become the fastest Quadro card on the market, replacing the Quadro P6000. The device is aimed primarily at virtual reality content creators, engineering applications and simulations, and is able to support displays up to 5120x2880 pixels at 60Hz.
The Quadro GP100 will have a 32-bit floating point performance of about 12 teraflops via 3,584 CUDA cores. It also features 64-bit floating point performance of 5 teraflops via 1,792 cores for precise scientific calculations. In addition, it uses 16GB of HBM2 (High Bandwith Memory) rather than the 24GB of DRR5X on the Quadro P6000, along with a two-way NVLink which allows GPUs to communicate faster than standard SLI configurations. This is the first time Windows systems will be able to use NVLink technology, as it has previously been confined to servers running Linux like IBM’s POWER8 and Nvidia’s GDX-1.
It will make its debut in Las Vegas at SOLIDWORKS World 2017 and is scheduled for release sometime in March. The GP100 will be available worldwide, but these powerful GPUs don’t come cheap. Pricing is not available yet for the GP100, but it will likely be in a similar ballpark as the Quadro P6000, which retails for several thousand dollars.