NVIDIA Announces New DGX SuperPOD, the First Cloud-Native, Multi-Client Supercomputer

0
68

NVIDIA introduced the first cloud-native, multi-client AI supercomputer: the NVIDIA DGX SuperPOD ™ next-generation processor features NVIDIA BlueField ® -2 DPUs.

By empowering the DGX SuperPOD with BlueField-2 DPUs (Data Processing Units that offload, accelerate and isolate user data), it is possible to provide customers with secure connections to their AI infrastructure.

The company also introduced NVIDIA Base Command ™ , which enables multiple users and IT teams to securely access, share and operate your DGX SuperPOD infrastructure. Base Command coordinates AI operations and training on the DGX SuperPOD infrastructure to enable the work of teams of data scientists and developers located around the world.

“AI is the most powerful technology the world has ever known, and NVIDIA DGX systems are the most powerful tool to harness it,” said Charlie Boyle, vice president and general manager of DGX systems at NVIDIA. “Combining multiple DGX systems, the new DGX SuperPOD provides an out-of-the-box AI data center that can be securely shared across entire teams of researchers and developers.”

Advance AI with DGX SuperPOD

DGX SuperPODs are AI supercomputers with 20 NVIDIA DGX A100 systems ™ or more and NVIDIA InfiniBand HDR networks. Some of the latest companies implementing DGX SuperPOD to power new AI solutions and services:

  • Sony Group Corporation is using DGX SuperPOD to enable its corporate research and development team to infuse AI throughout the company.
  • NAVER , a leading Internet technology company in Korea and Japan, is training giant AI language models at scale on DGX SuperPOD to pioneer new services in e-commerce, search, entertainment, and payment applications.
  • Recursion , a digital biology company working to industrialize drug discovery, is using DGX SuperPOD to accelerate its deep learning models and empower its growing workforce of machine learning experts.
  • MTS , Russia’s largest telecommunications company, is using its DGX SuperPOD as a foundation for its building AI-ready development infrastructure and new public cloud service offerings, as well as for integrating AI and data science into all of its departments.
  • VinAI is using its DGX SuperPOD, which will be the fastest AI supercomputer in Vietnam, to accelerate AI initiatives including autonomous vehicles, healthcare, and consumer services in Southeast Asia and around the world.

Additionally, NVIDIA and Schrödinger today separately announced a strategic partnership designed to leverage DGX SuperPODs to further accelerate supercomputer-scale drug discovery. Their co-developed solution can enable pharmaceutical and biotech companies of all sizes to simulate molecular combinations with physics and AI to identify and optimize the most promising compounds for potential therapeutic use.

Secure, Cloud-Native AI at Scale

NVIDIA designed the new DGX SuperPOD to meet customers’ growing security and scalability requirements as AI adoption broadens. More and more, enterprise IT departments need to support the work of multiple teams in different locations, while academic and research institutions often give external organizations access to their computing resources.

BlueField-2 DPUs offload, accelerate, and isolate users and their data, enabling organizations to securely provide private cloud access for their DGX SuperPOD infrastructure with security that spans user traffic, firewalls, and user access. multiple clients to storage. DDN is the first NVIDIA storage partner certified to support the next generation DGX SuperPOD with BlueField-2.

NVIDIA Base Command enables teams of AI developers and data scientists to seamlessly provision and schedule workloads on the DGX infrastructure, from prototyping to production.

To support the iterative process of building and refining AI models, Base Command provides built-in telemetry for users to validate deep learning techniques, workload configurations, and resource allocations to constantly improve outcomes.

New DGX Station A100 Subscription Makes AI More Accessible

NVIDIA also introduced a subscription offer available for the NVIDIA DGX Station A100 , the world’s only workgroup device in an office form factor that supports NVIDIA Multi-GPU Instance technology.

The new subscription program makes it easy for companies at every stage of growth to accelerate AI development external the data center for teams working in corporate offices, research facilities, labs and domestic offices.

Powered by four 80GB NVIDIA A100 GPUs, DGX Station A100 delivers 3X faster speeds compared to its predecessor for complex AI workloads such as natural language processing. Up to 28 data scientists can share one DGX Station A100 simultaneously, enabling companies to accelerate AI development and save up to $700,000 compared to data center servers.

Availability

Cloud-native, multi-client NVIDIA DGX SuperPODs will be available in Q2 2021 through NVIDIA global partners, who can provide pricing to qualified customers upon request.

NVIDIA Base Command will also be available starting in Q2.

Subscriptions for NVIDIA DGX Station A100 are available starting at a list price of $9,000 per month.

Leave A Reply

Please enter your comment!
Please enter your name here