- New Xeon 6 P-Core processors feature Priority Core Turbo (PCT) and Speed Select Technology (SST-TF).
- AI systems using the Xeon 6 CPUs can achieve up to 30% faster memory speeds and 20% more PCIe lanes for improved data transfer.
What happened: New Xeon 6 chips target AI bottlenecks
Intel has launched its much-anticipated Xeon 6 P-Core processors, designed specifically for GPU-powered AI systems. The new chips, including the Intel Xeon 6776P, feature innovations like Priority Core Turbo (PCT) and Speed Select Technology-Tuned Frequency (SST-TF), which aim to address CPU-side bottlenecks often encountered in high-performance AI environments.
These chips, part of the Granite Rapids family, are optimised for use in systems such as NVIDIA’s DGX B300, allowing GPUs to perform at their best without being held back by slower CPUs. The Xeon 6776P, a 64-core, 128-thread processor, promises to provide the raw power necessary for demanding AI workloads, facilitating faster processing and smoother performance.
Also Read: Intel reports $821M loss, despite modest Q1 performance
Also Read: Intel and Arqit boost quantum-safe security with cloud-based solution
Why it’s important
Intel’s new processors bring a significant leap in performance with two major enhancements. First, PCT increases the frequency of selected high-priority cores during demanding workloads, allowing AI models to run with minimal delay. Second, SST-TF adjusts core frequencies based on real-time load, ensuring that each core operates at peak efficiency.
The Xeon 6 chips also feature 30% faster memory speeds and an additional 20% more PCIe lanes compared to previous generations. These improvements directly address the growing demand for high-speed memory and faster data transfers between the CPU and GPU. Together, these advancements promise to enhance the efficiency of training and inference for large AI models, helping enterprises keep pace with the rapidly evolving AI landscape.
For businesses, Intel’s Xeon 6 processors offer a compelling solution to optimise GPU-accelerated workloads. With faster, more efficient processing, companies can expect to see performance boosts that will significantly improve productivity across various industries.