ARM Launches New Chip Design for HPC, Data Centers & Edge

Arm Unveils New Neoverse Platforms and Interconnect

Arm has recently announced the release of two new platforms, Arm Neoverse V1 and Neoverse N2, alongside a novel mesh interconnect designed for integration with these platforms. The V1 platform represents a completely new offering and exemplifies Arm’s growing aspirations within the data center, high-performance computing, and machine learning sectors.

Neoverse N2: Next-Generation General Compute

The N2 platform serves as Arm’s subsequent generation of general compute technology. It is engineered to support a diverse range of applications, extending from hyperscale cloud environments to SmartNICs and edge computing workloads. Notably, it is the inaugural design predicated on the company’s advanced Armv9 architecture.

Arm's Growing Presence in HPC

Historically, high-performance computing was largely controlled by a limited number of companies. However, the Arm ecosystem has achieved significant successes in this domain recently. Supercomputers located in South Korea, India, and France have chosen to utilize Arm technology.

The V1 platform is projected to deliver substantial performance improvements over the preceding N1 platform. These enhancements include a doubling of floating-point performance and a quadrupling of machine learning capabilities.

V1 Platform Details

“The primary focus of the V1 is maximizing performance – that was the core objective,” stated Chris Bergey, Senior Vice President and General Manager of Arm’s Infrastructure Line of Business. He further emphasized that V1 represents Arm’s broadest architecture to date.

While not exclusively designed for the HPC market, the high-performance computing sector was certainly a key target during development. Although the current Neoverse V1 platform does not yet incorporate the Armv9 architecture, future iterations will.

N2 Platform: Performance Per Watt

Conversely, the N2 platform prioritizes achieving optimal performance relative to power consumption, as Bergey highlighted. “Our aim is to maintain the performance-per-watt efficiency of the N1 while simultaneously increasing overall performance,” he explained. Testing conducted by Arm demonstrated a 1.3x performance gain for NGINX compared to the previous generation.

Customer Adoption and Industry Momentum

This launch also provides Arm with an opportunity to showcase its recent successes with customers. AWS Graviton2 has experienced considerable success, and Oracle is also leveraging Ampere’s Arm-based Altra CPUs for its cloud infrastructure.

“We are confident that Arm will become ubiquitous, spanning from the edge to the cloud,” said Bev Crair, Senior Vice President, Oracle Cloud Infrastructure Compute. “N1-based processors are consistently delivering the performance, scalability, and security that our customers demand from cloud infrastructure.”

Oracle, in collaboration with Ampere Computing and leading ISVs, is positioning Arm server-side development as a streamlined and cost-effective solution.

Expanding Ecosystem

Furthermore, Alibaba Cloud and Tencent are both investing in Arm-based hardware for their respective cloud services. Marvell will integrate the Neoverse V2 architecture into its OCTEON networking solutions.