Intel Xeon Scalable Chips Power Supercomputer With AI

San Diego Supercomputer Center (SDSC) at UC San Diego will be using Habana Labs’ artificial intelligence (AI) training and inference accelerators for Voyager, its supercomputer. This will allow the machine to leverage high-performance, high-efficiency AI computing for various applications.

Voyager will be used in advancing AI research across a range of science and engineering domains. And through the use of Habana’s accelerators, it will be able to efficiently scale AI training capacity with 336 Gaudi processors. They are meant for scaling large supercomputer training applications.

This is what Voyager will have:

• Supermicro X12 Gaudi AI Training System (SYS-420GH-TNGR) featuring eight Gaudi HL-205 cards paired with dual-socket 3rd Gen Intel® Xeon® Scalable processors, the newest high-performance processors from Intel, introduced this week
• Supermicro SuperServer 4029GP-T featuring eight Goya HL-100 PCIe cards for AI inference, paired with dual-socket 2nd Gen Intel® Xeon® scalable processors

  Habana says that Gaudi is the only AI processor to be able to natively integrate 10 100-gigabit ports of RoCE v2 (RDMA over Converged Ethernet) on-chip. But that’s not all, Voyager also features 16 Habana Goya processors to power AI inference mode. As such, data scientists and researchers can access machine learning frameworks and AI models.

“We are honored that Habana’s AI processors have been selected to power the AI workloads that will run on San Diego Supercomputer Center’s Voyager supercomputer,” said Eitan Medina, Chief Business Officer at Habana. “This implementation of our Gaudi and Goya products showcases how top academic institutions like SDSC can harness efficiency and performance.”

Voyager’s first three years of operation will be the Testbed Phase. During this time, SDSC will work with select research teams from astronomy, climate sciences, chemistry, particle physics and other fields to gain AI experience and insights.