As AI adoption accelerates, CPUs are essential to AI system operation as host nodes that manage core functions such as data preprocessing, transmission, and system orchestration. Intel continues to differentiate itself as the only semiconductor company to submit server CPU performance results to MLPerf.

Intel announced on the 2nd that it has proven the performance of Intel Xeon 6 P-Core in the latest MLPerf Inference v5.0 benchmark released by MLCommons.

Intel emphasized that the test results conducted on six major benchmarks showed that Xeon 6 improved AI performance by 1.9 times compared to the previous generation processor, proving that Xeon 6 is the optimal solution for the latest AI systems.

Intel Xeon 6 P-Core delivers up to 1.9x higher performance on average compared to 5th Gen Intel Xeon processors across key benchmarks including ResNet50, RetinaNet, 3D-UNet, and the new GNN-RGAT from MLPerf Inference v5.0.

Intel has been focusing on improving AI performance for the past four years, and since first submitting the 3rd generation Intel Xeon processor to MLPerf in 2021, ResNet50 performance has improved by a whopping 15x, and through software optimization, it achieved an additional 22% performance improvement in GPT-J and 11% in 3D U-Net.

The new MLPerf results are rolling out across OEM and ecosystem partner solutions, and as AI workloads become increasingly integrated with enterprise systems, OEMs are considering adopting Xeon-based systems to help their customers achieve the best AI performance.

“These MLPerf benchmark results demonstrate that Intel Xeon 6 is the CPU of choice for AI workloads, delivering the perfect balance of high performance and energy efficiency,” said Karin Epshitz-Segal, vice president and interim general manager of Intel’s Data Center and AI Group. “With continued performance improvements across a range of AI benchmarks, Intel Xeon continues to lead the market for CPUs for AI systems.”He explained, “It shows that they are underground.”

Meanwhile, Intel has collaborated with four major OEM partners, Cisco, Dell Technologies, Quanta, and Supermicro, to submit MLPerf results for Intel Xeon 6 P-Core, demonstrating various AI workloads and deployment capabilities.