Computex 2024: Intel Debuts Sierra Forest Xeon, Prices Gaudi AI Kits
In its Computex 2024 keynote address, Intel unveiled the sixth-generation Xeon processor, codenamed “Sierra Forest,” featuring an all-E-Core energy-efficient design, and announced pricing for its Gaudi 2 and Gaudi 3 artificial intelligence accelerator kits.
The sixth-generation Xeon processor, optimized for high-density, large-scale workloads in data centers, boasts significant improvements in performance and energy efficiency. Compared to the second-generation Xeon processor, it triples server cabinet density, reduces space requirements to one-third, and increases performance by up to 4.2 times, with a maximum per-watt performance improvement of 2.6 times.
Intel also revealed pricing for the previously announced Gaudi 2 and Gaudi 3 AI accelerator kits, which will be offered at $65,000 and $125,000 respectively. These prices are notably lower than competing products, making it more accessible for businesses to adopt AI capabilities in conjunction with Xeon processors.
Furthermore, Intel reiterated its collaboration with AMD, Broadcom, Cisco, Microsoft, Meta, and others in the Ultra Ethernet Consortium, establishing the UALink (Ultra Accelerator Link) connection standard to facilitate larger-scale server interconnection and enhanced AI computing performance.
Intel CEO Pat Gelsinger emphasized the company’s commitment to long-term partnerships with the Taiwanese supply chain, driving innovation from end devices to data centers and edge computing applications. He also highlighted Intel’s collaboration with Microsoft on the “Copilot + PC” product design ecosystem, accelerating the growth of the AI PC market.
In addition to unveiling the next-generation thin and light laptop processor, codenamed “Lunar Lake,” Gelsinger confirmed that the upcoming “Arrow Lake” processor will target desktops and high-performance devices. The “Panther Lake” processor, slated for release in 2025 and built on Intel’s 18A process, is expected to deliver even lower power consumption and higher computing performance.
