Intel has unveiled new hardware aimed at enhancing artificial intelligence (AI) workflows. The company introduced the Xeon 6 processor with upgraded Performance-cores (P-cores) and the Gaudi 3 AI Accelerator, specifically designed for enterprise customers and data centers. This new hardware promises increased throughput and cost efficiency, offering optimal performance per watt and a lower total cost of ownership. According to Intel, these innovations are designed to manage the growing demands of advanced AI models.
The Intel Xeon 6 features advanced Performance-cores and is tailored for data centers rather than retail consumers. It powers cloud servers for enterprises and boasts twice the performance of its predecessor, thanks to a higher core count. The processor also provides double the memory bandwidth and enhanced AI acceleration capabilities, enabling smooth support for large language models (LLMs). Intel claims that the Xeon 6 can meet AI performance needs across edge, data center, and cloud environments.
Intel’s Gaudi 3 AI Accelerator is the latest generation of specialized hardware designed to expedite AI tasks, particularly in deep learning, machine learning, and neural networks. Equipped with 64 Tensor cores, eight matrix multiplication engines (MMEs), 128GB of HBM2e memory, and 24 200GB Ethernet ports, it accelerates deep neural network computations and supports server scalability. Gaudi 3 is fully compatible with the PyTorch framework and advanced Hugging Face models, and Intel has partnered with IBM to deploy it for IBM Clouds, with Dell Technologies also utilizing the infrastructure.