AMD, a leading provider of electronic components, officially unveiled its latest data center and artificial intelligence (AI) technology products, including the MI300 series, at the "Data Center and AI Technology Premiere" event held in San Francisco, USA on June 14th, Beijing Time.
Dr. Lisa Su, Chairman and CEO of AMD, highlighted the significant market opportunities for AI, with the greatest potential coming from data centers. Last year, AMD introduced the groundbreaking 3D V-Cache technology, which formed the basis for the third-generation AMD EPYC Milan-X series processors.
Since then, 3D V-Cache has become a standard feature in EPYC processors. Today, AMD announced the launch of the fourth-generation EPYC processors, incorporating the updated 3D V-Cache technology.
The fourth-generation EPYC processors offer solutions for various scenarios and workload demands. They include "Genoa" for general computing, "Bergamo" for cloud-native computing (the industry's first x86 cloud-native CPU), "Genoa-X" for technical computing, and "Siena" for telecommunications/edge computing. The first three products have already begun shipping, while Siena is scheduled to be available in the second half of this year.
"Bergamo" features 128 "Zen 4c" cores and provides the highest vCPU density and industry-leading performance for applications running in the cloud. It is based on a specially customized Zen 4c microarchitecture, which maintains a similar feature set to Zen 4 microarchitecture while reducing core size requirements by half.
The "Bergamo" processor offers significant advantages, including a 2.7x energy-saving capability and a 3x capacity upgrade, enabling large-scale cloud-native applications. Its flexible scalability allows for a balance between core count and power consumption, driving improved performance.