Arm aims to be at the center of an increasingly diverse data center


Arm has been riding high in the mobile market for decades now, but has struggled to make its mark on servers. But the company hopes to reverse that trend with new initiatives that Arm executives discussed during the recent Arm DevSummit held last week.

The main initiative is to provide programming and design tools so that its chip designs can provide the fastest answers to solve complex problems without worrying about the hardware they are running on.

In the long term, Arm sees accelerators that include GPUs and neural processors will play an important role in computing, with processors based on the Armv9 architecture making it easier to distribute workloads across computers.

“In the modern world with increasingly diverse computing demands, we believe that specialized processing is the new standard bearer that will allow us to move beyond the faster, better and cheaper way of computing. general-purpose computing,” said Gary Campbell, executive vice president of central engineering at Arm, during a keynote that was streamed.

Arm has historically focused on its CPU technology, and 230 billion chips have shipped over the past 31 years in smartphones, in-vehicle devices and, to a lesser extent, PCs and servers. But Arm is now making inroads into traditional enterprise computing environments.

Developers want to write their application code and compile it to work without having to explicitly specify a chip to redirect a workload to, said Mark Hambleton, vice president of open source software.

In a keynote at Arm DevSummit, Hambleton said, “We observed from the server space that strict adherence to standards meant that operating systems would land on any server, from old to new, and would work, year after year. Arm servers already work this way.

This will result in time and cost savings, “faster innovation, better portability, and perhaps most importantly, it all just works,” Hambleton said in his keynote.

The concept of diverse computing environments with a multitude of chips is not new. Arm licensees, including Nvidia and Intel, have parallel programming software tools that automatically distribute computing to affected chips.

But having all the major known chipmakers as customers puts Arm in a bind because it can’t release its own specialized software framework or show a bias towards a specific customer.

As with Linux, Arm’s software efforts revolve around upstream code in the Linux kernel and working with open source consortia such as CNCF, Linaro, and Linux Foundation. Hardware customers can take Arm’s tools and adapt them to their needs. For example, Nvidia has placed its CPU roadmap directly on the Arm architecture, and it is modifying its CUDA parallel programming framework to accommodate its long-term Arm plans.

Source: Arm (September 2022)

Arm appears to be reversing a decade-long struggle to break into the server market, primarily thanks to Arm-based server chips offered by Ampere Computing. Oracle, Microsoft, and Google now offer virtual machine instances running on Arm processors for cloud-native applications. Amazon built the Graviton3 chip based on the Arm architecture.

Arm also offers coding tools through a program called Works on arm, which is now available from all major cloud providers and bare metal vendors such as Equinix. Developers can write and test applications specific to these computing environments. Arm also has its own virtual hardware environment in the cloud where developers can prototype applications before mass deployment.

The company’s data center roadmap includes high-performance computer chip designs called Neoverse V2, codenamed Demeter, and its successor, codenamed Poseidon, which will gradually support new technologies. The new chip designs will support DDR5 memory and PCIe Gen5 interconnects, but will differentiate on a technology called CXL (Compute Express Link), designed to accelerate parallel computing. Neoverse V2 will support CXL 2.0, while Poseidon will support CXL 3.0. Other Neoverse chips include the N-series chips for power efficiency and the E-series, which focuses on throughput.

The design of the Arm architecture relies on extensions that provide additional enhancement to handle data-intensive applications. One is SVE2, which is a vector extension available with an Armv9 architecture.

“SVE2 gives you greater ease of programming and portability. You can write and build software once for high portability on different Arm-based hardware with different SVE vector length implementations,” Campbell said.

Arm executives also talked about new security extensions in the area of ​​confidential computing, which involve firmware protection and a secure vault accessible only to authorized programs. Arm already has TrustZone extensions to protect sensitive data, but adds blocks to isolate business-critical data sets. Typically, organizations in key markets like finance and the military have protected datasets that drive AI models in enterprises.


Comments are closed.