Intel has strengthened its AI leadership by broadly supporting data centers and network environments with new Xeon 6 products equipped with high-performance P-cores that provide both performance and energy efficiency.

Intel announced today the launch of three new Intel® Xeon® 6 processors designed to support cutting-edge GPU-based AI systems.

This new product combines Priority Core Turbo (PCT) technology and Intel® Speed Select Technology-Turbo Frequency (Intel® SST-TF) functions to maximize CPU core frequency appropriately for AI workloads and further enhance GPU performance.

The new products will be available immediately, and in particular, the Intel® Xeon® 6776P has been adopted as the host CPU for NVIDIA's latest AI acceleration system, the DGX B300.

The Xeon 6776P in the DGX B300 provides the memory capacity and bandwidth required to operate large models and datasets, and plays a role in effectively tuning and managing the AI acceleration system.

“The new Xeon 6 lineup is the most optimized for next-generation GPU-accelerated AI systems,” said Karin Eibschitz Segal, vice president of Intel. “We will further strengthen our collaboration with NVIDIA to provide industry-leading AI systems and accelerate the adoption of AI,” he said.

PCT technology, combined with Intel SST-TF capabilities, dramatically improves AI workload performance.

Turbo frequency is dynamically allocated to high-priority HP cores to enable them to operate at higher speeds, while low-priority LP cores operate at the base frequency to efficiently distribute CPU resources.

This allows data to be fed to the GPU faster and maximizes the efficiency of the overall system.

The Xeon 6 P-Core processor delivers industry-leading performance for all AI systems, with up to 128 P-cores per CPU to balance complex AI workloads.

Memory speed is also improved by up to 30%, providing high memory bandwidth with MRDIMM and CXL support, and PCIe lanes are increased by up to 20% to optimize data transfer with fast I/O performance.

It maximizes system uptime with high reliability and ease of maintenance, and supports Intel Advanced Vector Extensions (AMX) to optimize AI data processing with FP16 operations.