Arm introduces Neoverse V1/N2, the newest chip architecture for high-performance computing and data centers

Arm also released two new systems, the Arm Neoverse V1 and Neoverse N2, as well as mesh interconnect technologies. V1 is a brand-new product, as the name suggests, and could be the best indication of Arm’s potential aspirations in the data center, high-performance computing, and deep learning. N2 is Arm’s next-generation general-purpose computing architecture, with support for anything from hyperscale clouds to smart network cards and edge workloads. It’s also the company’s first concept built on the latest Armv9 architecture.

High-performance computing used to be owned by a limited number of companies, but the Arm ecosystem has recently gained a large share of the market, and supercomputers in South Korea, India, and France are all betting on it. The promise of V1 is that it will outperform the old N1 platform in several ways. For example, floating-point performance will be doubled, and machine learning performance will be quadrupled.

“V1 is to prove how much performance we can deliver,” Chris Bergey, senior vice president and general manager of Arm’s infrastructure business line, said. He also said that V1 is Arm’s most comprehensive architecture yet. Although V1 It was not designed exclusively for the HPC industry, high-density computing is a clear goal. About the fact that the latest Neoverse V1 platform does not use the modern Armv9 architecture, the next generation would.

N2, on the other hand, emphasizes optimal output per watt, according to Bergey. “This is simply to keep the same envelope of output per watt that we had in N1, but add more performance,” he said. In the Arm comparison, for example, NGINX performed 1.3 times better than the previous generation product.

In several respects, today’s launch also serves as an occasion for Arm to showcase some of its latest customer wins. While AWS’ Graviton2 is obviously performing well, Oracle’s cloud architecture is built on Ampere’s Arm-based Altra CPU.

“From the edge to the cloud, we think Arm will be present. “We see that N1-based processors have the consistent consistency, scalability, and protection that customers demand from cloud infrastructure,” said Oracle Cloud Infrastructure Computing Senior Vice President Bev Crair. Oracle is making Arm server-side development a first-class, convenient, and cost-effective approach in collaboration with Ampere Computing and leading ISVs.”

Simultaneously, Alibaba Cloud and Tencent are investing in Arm-based hardware for their cloud services, and Marvell has confirmed that its OCTEON network solution will use the Neoverse V2 architecture.