On Wednesday, September 14, Arm held a press conference to announce the latest roadmap updates for its server processors, which the company calls “Neoverse” cores. The main reveal from the press conference was the V2 core (codenamed Demeter), which will power Nvidia’s “Grace” processor for high-performance computing (HPC) and artificial intelligence (AI) workloads.
The Arm Neoverse V, N, and E series cores target three different performance and power trade-offs. V-cores deliver higher performance per core at higher power and larger die area. E-cores are designed to be much more energy and die space efficient. N-cores sit between the other two core types with a balance of performance, scalability, and power. Arm has updated roadmaps for all three Neoverse processor core types with support for the latest platform technologies: PCIe (gen 5 and 6), CXL (2.0 and 3.0), and DDR5 memory.
Ampere Computing leads the way
Arm discussed several new design successes from key customers – including Microsoft Azure, Google Cloud – but these companies all use Ampere Altra products. Ampere Computing is the only processor company that currently offers merchant (independent) Arm server chips for cloud service providers.
For those unfamiliar with the company, it is a private start-up led by founder, president and CEO Renée J. James. She founded the company in 2017 after serving as president of Intel. She was also the first female general manager of an Intel business unit and the company’s first female executive vice president. She founded the company with a mission to fill the void in cloud computing and built a team that included veterans from Intel, Sun Micro and other large companies.
Ampere Computing has launched two products in the past two years: the 80-core Ampere Altra in 2020 and the 128-core Ampere Altra Max in 2021. The company has already tested its latest product, the AmpereOne, with customers earlier this year. While Ampere Computing used Arm’s Neoverse N1 cores to get its Altra line to market faster, the company’s real goal was to develop its own processor cores using only an Arm ISA license.
The Ampere Altra family has proven that it is possible to instantiate more CPU cores per socket while reducing power consumption by using cutting-edge 7nm processing technology compared to competing server CPU cores. This is a very attractive value proposition for cloud service providers because they can run more instances per rack per watt. The Ampere Altra Max packs 128 physical cores on a chip, and the performance of those cores scales linearly, as Ampere’s server chip design is optimized for cloud scaling using on-chip smart mesh network (NOC) and plenty of I/O and memory bandwidth. . Each processor core is designed with large cache and consistently fast clock speed. The chip’s power efficiency allows it to maintain consistent performance under heavy workloads.
Modern cloud infrastructure must be scalable and must extend seamlessly beyond the primary data center. Hybrid multi-cloud architectures require distributed resources from large-scale data centers, on-premises private clouds, edge resources, and all points in between. Ampere “Cloud Native” processing is optimized for such a distributed environment with consistent performance and low power consumption.
Cloud (and other) services running Ampere Altra (Max)
Arm-based cloud instances are making significant market share gains based largely on Ampere CPU chips. While interest in Arm-based cloud instances first surfaced with the first AWS Graviton instances released in 2018, the real growth has recently been driven by enterprises using Ampere Altra processors. Oracle Cloud was an early adopter of Ampere chips, but more recently, in September, Microsoft released new Azure VMs based on Ampere, and Google Cloud released a T2A preview instance in August, also based on Ampere processors. HPE announced an Ampere-based ProLiant system at HPE Discover in July, and European web hosting company Hetzner announced Ampere-based instances in August.
Low-power computing application based on Ampere server chips extends to autonomous driving – Cruise self-driving car development platform uses Altra processor to reduce power consumption while achieving computing performance required. Arm-based servers are also great for supporting Android cloud gaming. For example, Nvidia and Ampere are working together on a project called AICAN (Android-in-Cloud-with-Ampere-and-NVIDIA). The AICAN server platform uses Ampere Altra processors and NVIDIA GPUs that natively run Arm-compatible Android mobile games without modification or emulation. Another example: the ability to reduce operating costs while providing more cores allows Red Bull Racing to use Oracle Cloud and Ampere Altra for better and faster racing simulations.
The market share of Arm-based servers is growing, especially in cloud instances. While AWS has designed its own Arm-based Graviton processor, many other cloud service providers have relied on Ampere Altra products to pack more processor cores per rack and lower operating costs, without investing in a server processor design team. Ampere’s server chips give them all the benefits without the risks.
Arm’s new Neoverse roadmap, Alibaba Yitian 710, AWS Graviton 3, Nvidia’s Grace and Ampere’s AmpereOne lead to a robust Arm server ecosystem. While the Alibaba and AWS chips are designed strictly for internal use and the Nvidia Grace chip targets HPC and AI workloads, the Ampere Computing chips aim to bring predictable high performance and efficiency to the cloud and edge. of the network.
The Ampere Computing solution is set to get even more appealing with the AmpereOne, which uses a new custom processor core, based on the Arm instruction set but designed entirely by the company’s experienced design team and manufactured with advanced technology. 5 nm advanced process. Details of AmpereOne are yet to come. Meanwhile, Ampere Computing is just getting started in its plans to lead the development and deployment of cloud servers.
Update: New picture of Renée J. James
Tirias Research follows and advises companies across the entire electronics ecosystem, from semiconductors to systems and sensors to the cloud. Members of the Tirias research team consulted with AMD, Ampere, Arm, Intel, Nvidia, Qualcomm, Synopsys and other companies in cloud and IP ecosystems.