Arm Builds the Neoverse for Cloud, Edge Computing

At the company’s TechCon show, Arm officials unveil a roadmap for new compute platforms designed for a distributed world with trillions of connected devices.

Arm Neoverse

Arm officials are keeping a focus on the cloud and network edge at this week’s TechCon 2018 show as the industry prepares for the tens of billions of smart, connected devices on the horizon and the massive amounts of data those devices will produce.

The chip designer on Oct. 16 unveiled its plans for a family of infrastructure-class compute platforms aimed at hyperscale cloud data centers to the network edge that are playing an increasingly important role in a rapidly growing distributed IT computing environment driven by the rise of the internet of things (IoT). Arm officials also introduced a roadmap for its Neoverse solutions that will begin this year with the 16-nanometer Cosmos platform, continuing next year with the 7nm Ares offerings.

That will be followed by another 7nm+ platform, dubbed Zeus, in 2020 and then the 5nm Poseidon platform a year later. Each generation will bring a 30 percent performance gain and help drive innovation from a broad range of technology partners for systems and software.

Drew Henry, senior vice president and general manager of Arm’s Infrastructure Line of Business, said during an online press briefing before the start of the show that Cosmos will be a platform “derivative” that will include the company’s Cortex-A72 and Cortex-A75 64-bit architectures as well as infrastructure features.

Starting with Ares next year, the Neoverse lineup will become more of a pure platform offering with an architecture designed specifically for hyperscale data center-to-edge computing, with the Cortex architecture continuing to target the connected and intelligent devices.

The Neoverse IP is designed to offer the high performance, security and scalability needed as the number of connected devices proliferates and the amount of data needed to be collected, stored, processed and analyze grows. Arm-based systems-on-a-chip (SoCs) already power many of the devices that make up the IoT and play in such areas as networking.

“This puts us into a position to see how the internet will need to be architected to accommodate these trillion connected and intelligent devices,” Henry said. “There’s redistribution in the way processing is being done and processing is being moved from the core to the edge and sometimes to the devices themselves.”

Arm became the dominant chip designer for smartphones and tablets, with companies like Qualcomm and Samsung building their mobile SoCs based on Arm’s high-performance, low-power designs. Arm held off a push into the mobile chip space by Intel and over the past decade has looked to drive its architecture into the data center. The company has ramped up its efforts in the IoT since Japanese IT giant SoftBank bought Arm for $32 billion in 2016.

In August, Arm officials unveiled the company’s Pelion IoT platform, which is designed to give enterprises greater visibility into the connected devices on their networks, tools to optimize data traffic coming from the devices and the ability to manage it all in private, public or hybrid clouds or in on-premises environments.

Pelion Partners

During the first day of TechCon Oct. 15, Arm officials announced several new technology partners for Pelion, including rival Intel, as well as Arduino and myDevices.

“IoT requires a strong ecosystem of companies working together to deliver value, and one size does not fit all,” Dipesh Patel, president of Arm’s IoT Services Group, wrote in a blog post.” IoT takes a village, and Arm is tapping into its leadership in building ecosystems by fostering a diverse team of partners to enable IoT to scale securely.

Neoverse designs will include a wide range of Arm technology, including its TrustZone security architecture, CPU and system IP, as well as security and system IP from partners, industry standard technologies like NIST and Open Compute Project standards, and open-source software. There also will be an array of options for systems using the Neoverse chips, including accelerators like field-programmable gate arrays (FPGAs), memory such as high-bandwidth memory (HBM), storage-class memory (SCM), DDR and flash, and interconnects such as PCIe and 100 Gigabit Ethernet.

Scalability will be key in Neoverse, according to company officials. For networking, storage and security accelerated processing, the architecture can feature a 2x4 or 4x4 mesh with 16 to 24 CPUs and a 16MB cache for workloads like software-defined networking and storage and network-functions virtualization (NFV). In edge systems and jobs like virtual-network functions (VNF), there will be a 4x4 mesh, 32 CPUs and 32MB of cache, and for cloud computing environments running such applications as Kubernetes, containers and data analytics, the 1TB will be 8x8, between 48 and 96 CPUs and a 64MB cache.

Arm also listed a range of ecosystem partners, including chip makers such as Qualcomm, Samsung, Xilinx, NXP and Broadcom; cloud players like Microsoft Azure, Alibaba and Tencent; system makers including Hewlett Packard Enterprise, Cisco Systems, Nokia, Ericsson and Huawei; and operators like SoftBank, Sprint and Orange. Neoverse also is getting support from operating system makers such as Red Hat, Oracle and Canonical and container and virtualization vendors including Docker and VMware.