Intel Xeon 6 Processors and Gaudi 3 AI Accelerators With Ability to Handle Advanced AI Workloads Launched

 

Intel Xeon 6 Processors and Gaudi 3 AI Accelerators With Ability to Handle Advanced AI Workloads Launched

Intel recently unveiled new hardware focused at improving artificial intelligence (AI) workflows. The company introduced the Xeon 6 processor with new Performance-cores (P-cores) and Gaudi 3 AI Accelerator for enterprise customers and data centres on Tuesday. The chipmaker claims that the new hardware will offer both higher throughput and better cost optimisation to enable optimal performance per watt and lower total cost of ownership. These devices were launched to enable enterprises to handle the continuously increasing workload demands from more advanced AI models, according to the chipmaker.

Intel Xeon 6 Processor Launched

 The chipmaker says that its new Intel Xeon 6 is equipped with Performance-cores. These processors are not meant for retail consumers and instead will power data centres for enterprises to help them run cloud servers.

Intel claims the Xeon 6 processor offers twice the performance of its predecessor due to increased core count. It also offers double the memory bandwidth and AI acceleration capabilities.

Since it is hardware-based acceleration, it can run support very large language models (LLMs) with ease. It can “meet the performance demands of AI from edge to data centre and cloud environments,” according to Intel.

Intel Unveils Gaudi 3 AI Accelerators

Gaudi 3 is a new-generation AI Accelerator from Intel. These are specialised hardware chips designed to help machines in speeding up AI tasks, especially those related to deep learning, machine learning, and neural networks. These include GPUs, Application-Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), and Neural Processing Units (NPUs).

The Gaudi 3 AI Accelerator features 64 Tensor processor cores and eight matrix multiplication engines (MMEs) which are designed to accelerate deep neural network computations. It sports 128GB of HBM2e memory for training and inference, and 24 200GB Ethernet ports that enable scaling up the servers.

Intel's new AI Accelerator is compatible with the PyTorch framework and advanced Hugging Face transformer and diffuser models. The company has already tied up with IBM to deploy Gaudi 3 for IBM Clouds. Dell Technologies is also using the infrastructure for its data centres.