LOS ANGELES — Nvidia claims it has achieved a breakthrough in edge supercomputing and is working with Ericsson to build what it describes as the “world’s first software defined 5G RAN (radio access network).”

From the company’s vantage point at the intersection of computer graphics, high performance computing, and artificial intelligence (AI), Nvidia is releasing a CPU that is configured for myriad use cases off the shelf including 5G network architecture, IoT, and AI. The Nvidia EGX Edge Supercomputing Platform, which is cloud native and powered by the Cuda Tensor Core GPU, can process 15 teraflops of data per second and up to 140 simultaneous high-definition video streams.

“All of this processing basically translates to one single node [that] is equivalent to hundreds of nodes in a data center,” CEO Jensen Huang said during his keynote at MWC Los Angeles 2019.

Nvidia announced a series of partnerships for the EGX including: Red Hat to implement Kubernetes for a carrier-grade software stack; Ericsson to collaborate on the development of a virtualized 5G RAN; and Microsoft to integrate the company’s Azure cloud platform to advance edge-to-cloud AI computation.

The company has also earned new partnerships with data center server and software vendors, including Atos, Dell Technologies, Fujitsu, Mellanox, Lenovo, QCT, Super Micro, Cisco, and VMware.

Nvidia is targeting specific challenges that have vexed network operators for years. “The fundamental problem, of course, is that what used to be a phone, the cloud, and the telcos essentially being a pipe has to change,” Huang said. “The edge can no longer just be a pipe.”

That vision for the edge as a computing platform, which is shared by others in the industry, is driven by the need for low-latency computation that will power forward-looking applications in IoT, cloud gaming, AI-powered analytics, and SDN, Huang explained.

“The second fundamental problem is the amount of sensor data that is going to be streaming across this pipe is going to grow incredibly,” he said. “5G is going to make it possible for us to connect up to 1000 times more things than we currently do with 4G — from billions of things to trillions of things. Those things are going to be streaming sensor information that are going to be high density, high data rate, as well as continuous.”

Huang admitted that while there are trials in the works that demonstrate the combined capabilities of software-defined 5G networks and edge computing, most of those advancements are still under development and not yet commercialized at scale. In other words, new apps are coming but no one can say with certainty which services will eventually rise above and become widely adopted.

“My attitude about most of these things — because we're in the computer industry and we build computing platforms — is if you build it, they might not show up. But if you don't build it, they can't show up,” he said.

Huang is, however, confident that practically everything will eventually be connected to the cloud with a platform riding on top that powers imaginative services. Every company and industry will have an opportunity to dramatically change their business, he said.

Some of those advancements may seem small on the surface, but squeezing more energy out of the power grid, extracting more efficiency out of the transportation system, and getting more spectral efficiency out of wireless networks are gains that will be measured by trillions of dollars on an annual basis, according to Huang. “The impact of bringing intelligence out to the edge is extraordinary.”

Early instances of 5G are delivering higher bandwidth and spectral efficiency gains, but that only scratches the surface of what 5G promises to make possible, Huang explained. “The fact of the matter is, if we made all of our cell phone experiences better it would be a wonderful achievement but it certainly wouldn't be revolutionary,” he said.

Nvidia revealed some early apps and SDKs designed to showcase the capabilities of EGX including: Metropolis IoT for sensors in smart cities or retail environments; Jarvis Multi-Modal AI for pre-trained AI models for speech recognition and synthesis; and Isaac for the development of robotics in industrial applications.

Underlining the opportunities that have yet to be developed, Huang noted that there are 570 million farms covering 40% of the world’s land mass, 2 million factories, 13 million stores, 5 million call centers and 10 trillion miles driven every year across 40 million miles of roads. “We finally have the necessary technology to go engage” with these industries, he said.

But of all the announcement Nvidia made at its first appearance at MWC, the partnership with Ericsson on 5G virtualized RAN (vRAN) could be the most impactful and provide the company with a genuine entry point into telecom network architecture. “The world leader in wireless communications has been secretly working with us to create the world’s first software defined 5G RAN,” Huang said, adding that the companies aren’t ready to show the results of their work yet.

“The future is software defined, and these low-latency applications that have to be delivered at the edge can now be provisioned at the edge,” he said. “That future will become software defined high performance computing.”

Photo: Nvidia CEO Jensen Huang reveals the Nvidia EGX Edge Supercomputing Platform during his keynote at MWC Los Angeles 2019.