Nvidia is also deploying Base Command to provide its DGX SuperPOD as a service.
An extensive Nvidia certification program now includes servers with the company’s BlueField-2 data processing units (DPUs) and those with Arm processors. The graphics and data acceleration processing vendor is also set to deploy its DGX SuperPOD infrastructure as a service.
The company announced the extended certification and DGX SuperPOD as a service last week at the Computex virtual event. The two announcements are the latest steps taken by Nvidia to make it easier and more affordable to run applications with artificial intelligence. Because AI requires expensive and complex high-performance computing resources, it is beyond the reach of most organizations.
Manuvir Das, head of corporate IT at Nvidia, gave a keynote address at Computex, where he presented the company’s announcements.
“It’s time to democratize AI by bringing its transformative power to every business and its customers,” said Das.
To achieve this, Nvidia recently launched its Bluefield-2 data processing units (DPUs), software-programmable processors with high-performance network interfaces. Nvidia acquired its DPU technology with the acquisition of Mellanox for $ 7 billion last year. The company also activates servers based on Arm’s low power processors.
“We believe the opportunity for Nvidia in the enterprise market is substantial,” said Craig Weinstein, vice president of NVIDIA’s partner organization for the Americas. “In AI, there’s a lot of consulting work that’s required, depending on the industries we serve, and many clients in those industries need a lot of help.”
Asus, Dell Technologies, Gigabyte, QCT and Supermicro have launched servers with Nvidia’s BlueField-2 DPUs. As IT increasingly focuses on processing large amounts of data, today’s servers can’t handle everything, Das said.
“The flow of data to the network is becoming crucial to both the capacity and the security of the data center,” said Das. “A new kind of hardware is needed that sits in the data path and intelligently optimizes, inspects and protects data and protects applications from each other. Each server will need a DPU.
Nvidia’s DPUs support infrastructure tasks traditionally performed on processors, according to the company. This frees up capacity on the server processors to run applications, which NVIDIA says is more efficient.
Overall, Nvidia has added more than a dozen Certified System Partners, with the total now over 50. Some of the latest new Certified Partners include Dell, HPE, Nettrix, and Supermicro, which have launched servers based on the Nvidia’s HGX accelerated computing platform. Servers include 4 or 8 Nvidia A100 GPUs, Nvidia NVLink GPU interconnects, Nvidia InfiniBand network interfaces, and the company’s AI and HPC software stack.
Certified for New Arm servers
Nvidia has also added a certification program for Arm-based CPU servers, designed to provide high-performance computing to run AI-based workloads. Gigabyte and Wiwynn have said they will launch servers with Arm’s Neoverse processors and NVIDIA Ampere architecture GPUs or BlueField-2 DPUs. Once Nvidia certifies them, research the servers next year.
Gigabyte is teaming up with Nvidia to create an Arm HPC development kit for programmers to build AI and scientific computing applications. The kit comes with an Arm processor and Nvidia’s A100 Tensor Core GPU server APIs.
While approval of Nvidia’s ongoing $ 40 billion deal to acquire Arm remains uncertain, it will not impact the new certification program, Das said.
Basic ordering for SuperPOD as a service
Nvidia’s new supercomputing infrastructure as a service DGX SuperPOD will arrive this summer. To provide this capability, Nvidia has launched Base Command, which will make the DGX SuperPOD available in smaller increments.
The minimum configuration for a DGX SuperPOD is a cluster of 20 DGX systems.
“The pinnacle of AI capability is the DGX SuperPod,” Das said. “The first step in democratization is to make this cutting-edge machine more accessible and more accessible. “
With Base Command, customers will be able to use as little as 3 DGX systems for just a few months.
Base Command, a Kubernetes-based software stack, “allows administrators to share this powerful supercomputer between an organization of data scientists and a mix of workloads,” said Dais. Like DGX POD, SuperPOD is a supercomputer that uses NetApp’s ONTAP AI data management platform, he added.
As a hosted offering, Nvidia will manage the infrastructure. The offering includes all NetApp flash storage and data management. In addition, it is available in Equinix data centers. Nvidia announced that it is also working with AWS and Google to offer Base Command from their respective cloud services.
“This work holds the promise of a true hybrid AI experience for customers – write once, run anywhere,” Das said.
The subscription service is available for early access now; the company expects more availability this summer. The monthly subscription price starts at $ 90,000.