Arm Unveils First 64-Bit Real-Time Processor: Comes With Computational Storage

Although Arm’s chip division may be sold, Arm’s technicians have not stopped.

Arm this week unveiled its first 64-bit real-time processor, the latest addition to its Cortex-R family, which includes Linux support and an emphasis on enterprise storage applications. This strategy reflects the growing need to bring processing and analysis closer to the data.

The 64-bit Cortex-R82 adds “computational storage” capabilities such as support for Linux microservices and cloud-native software development with memory management options. It also addresses the need to address larger memory capacities, which the company says is becoming a barrier to performance when running Linux or emerging computing storage devices.

“Computational storage is all about making storage devices smarter to process data directly where it’s stored,” Arm noted.

Based on workload testing, the Cortex-R82 doubles the performance of the previous Cortex-R5 and -R8, while also integrating the company’s strengths in traditional enterprise storage infrastructure, including drive controllers. Traditionally, storage controllers run bare metal workloads. To extend these capabilities, the Cortex-R82 includes an optional memory management unit, which allows the operating system to run directly on the memory controller.

This opens up “opportunities for new and improved applications” to run on Linux-based microservices, said Neil Werdmuller, director of storage solutions at Arm. The company said the real-time processor also offers its Neon technology for accelerating machine learning workloads that take up many of the computational storage needs in enterprise data centers.

The R82 processor is also designed to address performance shortfalls such as memory access barriers and the growing need for businesses to run Linux-based services. “There is a need for higher performance, real-time computing with higher addressable space, and the ability to run Linux to support next-generation computing storage devices,” Werdmuller noted in a blog post.

In addition to Linux support, real-time processors can access up to 1Tb of DRAM address space while reallocating computing resources for different workloads. Meanwhile, the optional memory management unit supports Linux and other high-end operating systems as well as application code and a long list of database applications.

“The ability to run Linux provides developers with an entirely new set of software tools and technologies, such as Docker and Kubernetes, for their storage applications, providing an accelerated method of implementation,” Werdmuller added.

Arm is rolling out a growing number of chips, emphasizing that local data processing is a way to reduce latency by eliminating large file movements and other architectural tweaks. The benefits of this shift include database acceleration, faster data analysis, and faster applications, such as image and video processing for streaming applications.

Arm’s computational storage approach is designed to simplify data storage, retrieval and processing by replacing traditional data calls from storage with simple “operation” requests. This front-end step eliminates the need to move data around without processing operational requests in storage before delivering the desired results.

Arm said the approach is designed to address the growing number of data-driven workloads, such as machine learning models, that exceed the capacity of current data center servers. At the same time, Linux support expands the range of “drive” workloads.

“Running Linux and related software tools directly on the drives facilitates computational workload management and file system identification to perform on-drive computations and generate insights about the drives, greatly reducing data movement, latency and performance,” Werdmuller said. consumption.”

Flexible storage controllers also enable balancing, allowing different workload types running on the controller to be adjusted as needed. For example, the R-82 will facilitate real-time storage work during peak demand and then switch to data analysis tasks during off-peak hours.

Arm said the Cortex-R82 quad-core cluster is built on a 5nm process technology. The cluster configuration integrates 1Mb of L2 shared cache. The maximum clock frequency exceeds 1.8 GHz, with performance up to 5.82 MHz according to the CoreMark CPU and Embedded Microcontroller benchmarks.

It is understood that the Linux OS distributions supported by Cortex-R82 include Red Hat Enterprise Linus, SUSE and Ubuntu.


The Links:   CM150RX1-24T SKKT71/16D