Enhanced Nvidia Certification Program includes DPU, Arm Systems

Nvidia is also deploying Base Command to offer DGX SuperPOD as a service.

expanded Nvidia Certification programs now include company servers BlueField-2 A unit with a data processing unit (DPU) and an Arm-based processor. Graphics and data acceleration processing providers also DGX SuperPOD Infrastructure as a service.

Last week, the company announced extended certification and DGX SuperPOD as services. Computex Virtual event. Both announcements are the latest milestones by Nvidia to make it easier and more affordable to run applications using artificial intelligence. AI is out of reach of most organizations because it requires expensive, complex, high-performance computing resources.

Nvidia Manuvir Das on Computex 2021

Manuvir Das, Head of Enterprise Computing at Nvidia, gave a keynote speech at Computex to outline the company’s announcement.

“It’s time to democratize AI by bringing the transformational power of AI to all companies and their customers,” says Das.

To achieve that, Nvidia recently launched the Bluefield-2. Data processing unit (DPU)A software-programmable CPU with a high-performance network interface. Nvidia won last year’s DPU technology $ 7 billion acquisition Of Mellanox. The company also enables servers based on Arm’s low-power CPUs.

Craig Weinstein, vice president of NVIDIA’s Americas partner organization, said: “In the area of ​​AI, depending on the industry in which we serve, we need a lot of consulting work, and many customers in those industries need a lot of support.”

Asus, Dell Technologies, Gigabyte, QCT and Supermicro have launched servers with Nvidia’s BlueField-2 DPU. Today’s servers cannot process all data, as computing is driven by processing larger and larger amounts of data, Das said.

“The flow of data to the network is very important to both the functionality and security of the data center,” says Das. “We need a new kind of hardware that is on the data path, intelligently optimizing, inspecting, protecting data, and protecting applications from each other. Every server requires a DPU.”

According to the company, Nvidia’s DPU takes on infrastructure tasks that were traditionally performed on CPUs. This frees up space on the server CPU and allows you to run applications. This is what NVIDIA thinks is more efficient.

Overall, Nvidia has added more than 12 certified system partners, for a total of more than 50. The latest new certified partners include Dell, HPE, Nettrix and Supermicro. Nvidia HGX Accelerated computing platform. The server contains 4 or 8 Nvidia A100 GPUs. Nvidia NVLink GPU interconnect, Nvidia InfiniBand Network interface and company AI and HPC software stack.

Certification of new Arm server

Nvidia has also added a certification program for Arm-based CPU servers designed to provide high-performance computing for running AI-based workloads. Gigabyte And Wiwynn He said he would launch a server with Arm’s Neoverse-based CPU and either an NVIDIA Ampere architecture GPU or a BlueField-2DPU. Once Nvidia authenticates them, look for a server sometime next year.

Gigabyte has partnered with Nvidia to create Arm HPC Developer Kit Allows programmers to build AI and scientific computing applications. The kit includes an Arm CPU and Nvidia’s A100 Tensor Core GPU server API.

Nvidia approval pending $ 40 billion agreement The acquisition of Arm remains uncertain and will not affect the new certification program, Das said.

SuperPOD basic commands as a service

Nvidia’s new DGX SuperPOD supercomputing infrastructure as a service will arrive this summer. To provide that functionality, Nvidia has launched Base Command. This will gradually make the DGX SuperPOD available.

The minimum configuration of the DGX SuperPOD is a cluster of 20 DGX systems.

“The pinnacle of AI capabilities is the DGX SuperPod,” says Das. “The first step towards democratization is to make this best machine more accessible and accessible.”

With Base Command, customers will be able to use three DGX systems in just a few months.

With BaseCommand, a Kubernetes-based software stack, “administrators can share this powerful supercomputer throughout the data scientist’s organization and share different workloads,” Dais said. Like the DGX POD, he added that the SuperPOD is a supercomputer that uses NetApp’s ONTAP AI data management platform.

As a hosted offering, Nvidia manages the infrastructure. This offering includes all of NetApp’s flash storage and data management. It is also available in the Equinix data center. Nvidia also announced that it will work with AWS and Google to provide BaseCommand from their respective cloud services.

“This task promises our customers a true hybrid AI experience. Once written, it can be done anywhere,” says Das.

Subscription services are now available for early access. The company plans wider availability this summer. Monthly subscriptions start at $ 90,000.

Enhanced Nvidia Certification Program includes DPU, Arm Systems

Source link Enhanced Nvidia Certification Program includes DPU, Arm Systems

Related Articles

Back to top button