Intel is gearing up to launch its next-generation X86 processors specifically designed for artificial intelligence (AI) tasks, although these chips will not directly handle AI workloads.
During Computex this week, Intel introduced its Xeon 6 processor line, highlighting the Efficient-cores (E-cores) that promise up to 4.2 times the performance of the Xeon 5 processors. The initial Xeon 6 CPU, the Sierra Forest version (6700 series), is more performance-focused, with the Granite Rapids line featuring Performance cores (P-cores or 6900 series) set to debut next quarter.
The upgraded Xeon processors offer the potential for 3:1 data center rack consolidation while maintaining performance levels and achieving up to 2.6 times performance-per-watt improvements over their predecessors, as per Intel’s claims.
Reece Hayden, a principal analyst at ABI Research, emphasized the importance of energy efficiency in the data center AI market, particularly in terms of power consumption and environmental impact. Hayden noted that the Intel Xeon 6 processors will serve as the CPU head node in Gaudi-powered AI systems, contributing to reduced power consumption and a more sustainable energy footprint for AI technologies.
Increased rack density resulting from data center consolidation creates space for deploying AI-specific hardware to support training and inferencing, Hayden added.
An employee at the Intel Assembly Test facility in Kulim, Malaysia, inspecting Intel Xeon 6 Sierra Forest processors with E-Cores.
Intel Corp.
Intel also unveiled its Lunar Lake client processor line targeted at the AI PC sector. These x86 chips boast up to 40% lower system-on-chip (SoC) power consumption compared to the previous generation, offering enhanced efficiency for AI applications.
2024-06-10 10:00:02
Source from www.computerworld.com