Intel Unveils New Xeon 6 P-Core CPUs for Advanced AI Systems, Debuts in Nvidia DGX B300

Intel has recently unveiled three new Xeon 6 P-Core CPUs, specifically designed to meet the demands of advanced GPU-powered AI systems. These processors will make their debut in Nvidia's DGX B300 AI systems, which are at the forefront of AI acceleration technology. The new Xeon 6 P-Core CPUs are equipped with Intel's Performance-cores (P-cores) and introduce two innovative technologies: Intel Priority Core Turbo (PCT) and Intel Speed Select Technology – Turbo Frequency (SST-TF). Together, these technologies aim to enhance the performance of AI workloads by providing customizable core frequencies and optimized resource distribution. PCT allows for the dynamic prioritization of high-performance cores, enabling them to run at higher turbo frequencies while the lower-priority cores operate at base frequency. This setup ensures that critical tasks get the necessary processing power without overburdening the system. Intel states that PCT can run up to eight high-priority cores at elevated frequencies, improving overall efficiency and performance. In addition to PCT, the new Xeon 6 processors boast an impressive 128 P-cores per CPU, a 20% increase in PCIe lanes compared to the previous generation, with up to 192 lanes for a 2S server. Intel claims that its latest Xeon 6 CPUs offer 30% faster memory speeds than those of its competitors, particularly the latest AMD EPYC processors, thanks to Multiplexed Rank DIMMs (MRDIMMs) and Compute Express Link. These enhancements result in up to 2.3 times higher memory bandwidth compared to the previous generation, which is crucial for handling large datasets and complex AI models. The Xeon 6 processors support a 2 DIMMs per channel (2DPC) configuration, allowing for up to 8TB of system memory. MRDIMMs not only boost bandwidth and performance but also reduce latency, contributing to faster data retrieval. Each of the new CPUs features up to 504 MB of L3 cache, further accelerating data access. Intel's Advanced Matrix Extensions (AMX) technology, present in the Xeon 6 processors, can offload certain tasks to the CPU, enhancing efficiency in AI workloads. Intel has confirmed that AMX now supports FP16 precision arithmetic, which is essential for data pre-processing and other critical CPU tasks in AI applications. Here’s a breakdown of the specifications for the new Intel Xeon 6 P-Core CPUs: Intel Xeon 6732P Processor: 32 cores, 4.3 GHz max turbo frequency, 3.8 GHz base frequency, 144 MB cache, 350 W TDP. Intel Xeon 6774P Processor: 64 cores, 3.90 GHz max turbo frequency, 2.50 GHz base frequency, 336 MB cache, 350 W TDP. Intel Xeon 6776P Processor: 64 cores, 3.90 GHz max turbo frequency, 2.30 GHz base frequency, 336 MB cache, 350 W TDP. Intel Xeon 6716P-B Processor: 40 cores, 3.5 GHz max turbo frequency, 2.5 GHz base frequency, 160 MB cache, 235 W TDP. The Xeon 6716P-B variant stands out for its reduced power draw, making it a suitable option for environments that require balanced performance and energy efficiency. Intel emphasizes that its Xeon 6 processors with P-cores are well-suited to handle the growing complexities and demands of AI computing, offering an ideal blend of performance and efficiency. This announcement comes during Computex 2025, where Intel also introduced its new $299 aRc Pro B50 with 16GB of memory and 'Project Battlematrix' workstations equipped with 24GB Arc Pro 60 GPUs. These additional products highlight Intel's commitment to advancing both hardware and software solutions for AI and high-performance computing. For the latest updates, reviews, and analyses on Intel's new releases, follow Tom's Hardware on Google News and make sure to click the Follow button.