Oracle is partnering with Nvidia to offer a new AI supercomputing service, dubbed DGX Cloud and available immediately, using Oracle Cloud Infrastructure's Supercluster.
“OCI has excellent performance. They have a two-tier computing fabric and management network," Nvidia CEO Jensen Huang said during his keynote at the company’s annual GTC conference on Tuesday.
Nvidia is working with other cloud providers to provide similar services, but Oracle is its first partner to go live with an offering.
"Nvidia's CX7 along with Oracle’s non-blocking remote direct access memory (RDMA) forms the computing fabric," Huang said. "And Bluefield 3 will be the infrastructure processor for the management network. The combination is a state-of-the-art DGX AI supercomputer that can be offered as a multitenant cloud service.”
The announcement comes after Oracle Chairman Larry Ellison told analysts on a recent quarterly earnings conference call that the company has been experiencing increasing demand from enterprises running AI workloads.
OCI’s Supercluster includes OCI Compute Bare Metal, which provides an ultralow-latency RDMA over Converged Ethernet (RoCE) cluster based on networking technology, and a choice of high performance computing storage options.
AI service supports massively parallel applications
The AI supercomputing service, according to Oracle, can support thousands of OCI Compute Bare Metal instances with tens of thousands of Nvidia A100 GPUs for processing massively parallel applications.
“OCI Supercluster networking can now scale up to 4,096 OCI Compute Bare Metal instances with 32,768 A100 GPUs,” said Karan Batta, vice president of product at Oracle. She added that, “OCI plans to offer Compute Bare Metal instances with Nvidia H100 GPUs later this year.”
Nvidia had first announced its intent to launch the DGX Cloud Service in partnership with cloud service providers in the beginning of March when it announced quarterly earnings.
The DGX Cloud also will be supported by Microsoft, Google Cloud, and other cloud providers, Nvidia said.
Microsoft Azure is expected to begin hosting DGX Cloud next quarter, and the service will soon expand to Google Cloud and other providers, Nvidia said in its announcement Tuesday.