Intel says that the Gaudi 3 AI accelerator will deliver better performance than NVIDIA’s H100 — at a fraction of the cost
At Intel Vision 2024, the company detailed its broad enterprise AI strategy that emphasizes open, scalable systems that work across all AI segments. The announcements include a new chip called the Gaudi 3 and the company’s plans to create an open platform for enterprise generative AI (GenAI) applications in partnership with SAP, RedHat, VMware and others.
According to Intel, its Gaudi 3 AI accelerator will deliver 50% on average better interference and 40% on average better power efficiency across large-language-models (LLMs) than NVIDIA’s H100. And it will do this, claimed the company, “at a fraction of the cost.” Further, Intel’s accelerator is projected to deliver on average versus NVIDIA H100 50% faster time-to-train across Llama2 7B and 13B parameters and GPT-3 175B parameter models.
While this direct comparison stands out, it’s worth noting that NVIDIA announced its flagship B200 Blackwell chip last month, which it said can perform certain tasks 30-times faster than the H100. This chip is expected to hit the market later this year.
“Intel Gaudi 3 provides open, community-based software and industry-standard Ethernet networking. And it allows enterprises to scale flexibly from a single node to clusters, super-clusters and mega-clusters with thousands of nodes, supporting inference, fine-tuning and training at the largest scale,” stated the company.
The new chip will be available to original equipment manufacturers (OEMs) — including Dell Technologies, Hewlett Packard Enterprise, Lenovo and Supermicro — in Q3 of this year.
“In the ever-evolving landscape of the AI market, a significant gap persists in the current offerings. Feedback from our customers and the broader market underscores a desire for increased choice. Enterprises weigh considerations such as availability, scalability, performance, cost, and energy efficiency,” commented Justin Hotard, Intel executive vice president and general manager of the Data Center and AI Group. “Intel Gaudi 3 stands out as the GenAI alternative presenting a compelling combination of price performance, system scalability, and time-to-value advantage.”
Intel also announced collaborations with Google Cloud, Thales and Cohesity to leverage its confidential computing capabilities in their cloud instances.
