The new line of CPUs with high-performance cores and technologies like Priority Core Turbo is designed to maximize performance in artificial intelligence workloads. NVIDIA is already integrating them into its new DGX B300 systems.
Santa Clara, California, May 22, 2025 — Intel has introduced three new models of the Intel® Xeon® 6 processor series with performance cores (P-cores), specifically designed to manage and enhance GPU-accelerated artificial intelligence systems. Among the technological innovations are the additions of Priority Core Turbo (PCT) and Intel® Speed Select Technology – Turbo Frequency (SST-TF), two innovations that enable dynamic and customized management of CPU resources to optimize performance in AI environments.
One of these new processors, the Intel® Xeon® 6776P, is already in production and has been chosen as the host CPU for the NVIDIA DGX B300, NVIDIA’s new AI-accelerated computing system. This processor not only coordinates system tasks but also provides high memory capacity and bandwidth, critical for training and operating large-scale models.
A CPU Designed to Accelerate Artificial Intelligence
Karin Eibschitz Segal, Corporate Vice President and Acting General Manager of Intel’s Data Center Group, emphasized that “these new Xeon models showcase the unmatched performance of the Xeon 6 series, making them the ideal CPU for next-generation AI systems”.
The key to the performance leap lies in Priority Core Turbo (PCT), which allows for the dynamic allocation of priority cores to operate at higher turbo frequencies, while lower-priority cores maintain base frequencies. This allocation is especially useful for AI workloads requiring serial processing or sequential tasks, enhancing data feeding to GPUs and maximizing overall system efficiency.
Key Technical Features of the New Xeon 6
- High Core Density: Up to 128 high-performance (P-core) cores per processor, ideal for distributing demanding workloads.
- 30% Higher Memory Speed compared to competitors, compatible with MRDIMMs and advanced memory links like Compute Express Link (CXL).
- Improvements in I/O Performance: Up to 20% more PCIe lanes than previous generations, facilitating data access for intensive workloads.
- High Reliability and Maintainability: Designed with advanced availability and service features to maximize uptime.
- Support for Intel® Advanced Matrix Extensions (AMX): Enables FP16 precision calculations, accelerating critical preprocessing and computing tasks on CPUs for AI workflows.
A Commitment to Energy Efficiency and Scalability
Beyond raw performance, Intel emphasizes that its new Xeon 6 processors are designed to offer an optimal balance between performance and energy efficiency, adapting to both data center environments and network deployments. This evolution reinforces Intel’s commitment to becoming a leader in CPU solutions optimized for AI at a time when many companies are modernizing their infrastructure to meet the exponential growth of generative AI models.
The immediate availability of these processors, along with their implementation in leading systems like the NVIDIA DGX B300, represents a strategic step for Intel in the competitive market for high-performance AI infrastructure.