Nvidia developed space and power-efficient deep learning systems reducing physical footprint of AI clusters.

At its big annual Silicon Valley conference Tuesday, Nvidia unveiled what it said was the world's first single server with enough computing muscle to deliver two petaflops, a level of performance usually delivered by hundreds of servers networked into clusters. Aimed primarily at deep learning applications, the DGX-2 system is 10 times more powerful than the Volta GPU-powered version of its predecessor DGX-1, which was only released in September, according to the chipmaker. Nvidia achieved this performance by packing the DGX-2 with twice the amount of GPUs, upgrading each GPU with twice as much memory as before, using a brand new GPU interconnection technology, and improving Nvidia deep learning software. The Santa Clara, California-based company is the leading maker of GPUs, which accelerate computer graphics but also run inside machines used to train neural networks—the type of computing system that learns on its own by analyzing vast amounts of data and powers the AI boom.

The DGX-2 is the first system to use Nvidia's new GPU interconnection technology called NVSwitch, also announced at the GTC summit in San Jose. NVSwitch improves upon the previously existing GPU fabric NVLink by linking every one of the 16 GPUs inside the box to every other. According to Nvidia, it provides five times more bandwidth than the top PCIe switch on the market. "Every single GPU can communicate to every other single GPU at 20 times the bandwidth of PCI Express," Jensen Huang, Nvidia founder and CEO, said in his GTC keynote. The switch transfers 300GB per second. At the summit, the company also announced that its top-shelf data center GPU, the Tesla V100, now ships with double the memory—32GB—which power the new supercomputer in a box.

Nvidia priced the DGX-2 at $399,000 and positioned it as a major cost and space saver for data centers. A single DGX-2 box provides performance equivalent to "300 [Intel] Skylake servers," according to Jim McHugh, Nvidia's VP and general manager for Deep Learning Systems. McHugh noted that 300 Skylake servers represent "basically 15 racks of servers, which will save a lot" of space and power in the data center. Huang noted that the equivalent Skylake cluster would cost "easily for $3 million."

GPU sales have driven substantial revenue for Nvidia, with GPUs bringing $8.14 billion in revenue in the company's fiscal 2018. Of that, $1.93 billion came from GPUs sold into data centers for deep learning and more traditional high-performance computing workloads—a 133 percent increase from the previous fiscal year. Cloud giants like Amazon Web Services, Microsoft Azure, and Google Cloud Platform are responsible for much of Nvidia's data center GPU revenue, with all top cloud service providers deploying its GPUs to power their own AI applications and provide GPU computing power as a service.