Intel Corp. (Intel) has revealed its latest innovation, the Intel Gaudi 3 AI accelerator, at the Intel Vision event.
Intel enhanced the Gaudi 3 AI, including a quadrupling of AI compute for BF16, a 1.5x increase in memory bandwidth, and a doubling of networking bandwidth, promising to facilitate massive system scalability.
“Enterprises weigh considerations such as availability, scalability, performance, cost, and energy efficiency,” said 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.”
READ:
Lenovo beefs up Legion gaming laptops with Intel processors
3rd Gen Intel Xeon Scalable equipped with built-in AI, security
Enterprises across various sectors, including finance, manufacturing, and healthcare, are rapidly embracing AI technologies. To facilitate this transition effectively, businesses require open, cost-effective, and energy-efficient solutions. The Intel Gaudi 3 accelerator aims to meet these needs by offering versatility through open community-based software and industry-standard Ethernet networking, enabling businesses to scale their AI systems and applications flexibly.
GenAI
The Intel Gaudi 3 accelerator, designed for efficient large-scale AI compute, is manufactured using a 5-nanometer process. Its architecture allows for parallel activation of all engines, including the Matrix Multiplication Engine (MME), Tensor Processor Cores (TPCs), and Networking Interface Cards (NICs), facilitating fast and efficient deep learning computation and scale.
Key features of the Intel Gaudi 3 accelerator include a dedicated AI compute engine, memory boost for LLM capacity requirements, efficient system scaling for enterprise GenAI, and open industry software for developer productivity. Intel also introduces the Gaudi 3 PCIe add-in card, tailored for workloads such as fine-tuning, inference, and retrieval-augmented generation (RAG).
The Intel Gaudi 3 accelerator is projected to deliver significant performance improvements for training and inference tasks on leading GenAI models compared to competitors, including Nvidia. It will be available to original equipment manufacturers (OEMs) in the second quarter of 2024, with general availability expected in the third quarter. Notable OEM adopters include Dell Technologies, Hewlett Packard Enterprise, Lenovo, and Supermicro.
The Intel Gaudi 3 accelerator will power cost-effective cloud LLM infrastructures for training and inference, offering organizations price-performance advantages and choices. Among the notable adopters is NAVER, a leading technology company.