Arm’s ambition to take on the x86 CPUs in HPC, AI and specialized computing workloads took expression today with the introduction of Armv9, the first new Arm architecture in a decade. The launch comes on the heals of AMD’s March 16 introduction of 7nm EPYC “Milan” CPUs and ahead of new Intel 3rd Gen Xeon CPUs coming out on April 6.
Armv9 incorporates Scalable Vector Extension (SVE), a technology that Arm developed in partnership with Fujitsu and that is incorporated into Fugaku, the world’s fastest supercomputer. Building on that work, Arm said it has developed SVE2 for Armv9 to enable machine learning (ML) and digital signal processing (DSP) capabilities across a wider range of applications.
Vectors, one-dimensional arrays of data, came to HPC in 1976, with the Cray-1 supercomputer. Noting that the more vectors a computer can handle in parallel and the longer those vectors are the more powerful the computer will be – Arm said that while Armv8-A’s vectors are 128 bits in length, Armv9’s SVE2 upgrade lets chip designers choose a vector length in multiples of 128, up to 2048 bits. “That’s an enormous amount of parallel compute, and while SVE was initially developed for the HPC space, SVE2 in Armv9 extends SVE support for a range of specialized DSP and XR (augmented and virtual reality) workloads, from genomics to computer vision,” Arm said.
“SVE2 enhances the processing ability of 5G systems, virtual and augmented reality, and ML workloads running locally on CPUs, such as image processing and smart home applications,” Arm said in its announcement. “Over the next few years, Arm will further extend the AI capabilities of its technology with substantial enhancements in matrix multiplication within the CPU, in addition to ongoing AI innovations in its Mali GPUs and Ethos NPUs.”
Arm Ltd., a semiconductor and software design company owned by Softbank and based in the U.K., is the object of an $40 billion acquisition effort announced last September by Nvidia. It’s a potentially powerful combination now working its way through an extended regulatory review in the U.S., U.K. and China. If approved, the acquisition would give Nvidia a CPU to be tightly integrated for HPC and AI workloads with its industry leading GPU chips, just as AMD has its EPYC CPUs and MI100 GPUs and Intel will, possibly beginning next year, marry its “Ponte Vecchio” GPU to its advanced Xeon chips.
In its announcement, Arm claimed that over the past five years it has increased CPU performance annually at a rate outpacing the industry, and said it expects the Armv9 performance increases of more than 30 percent over the next two generations of mobile and infrastructure CPUs. This the company attributed to its Total Compute design methodology, which Arm said it will apply across its IP portfolio of automotive, client, infrastructure and IoT solutions. Arm said it also is plans to increase frequency, bandwidth and cache size, and reduce memory latency.
“Addressing the demand for more complex AI-based workloads is driving the need for more secure and specialized processing, which will be the key to unlocking new markets and opportunities,” said Richard Grisenthwaite, SVP, chief architect and fellow, Arm. “Armv9 will enable developers to build and program the trusted compute platforms of tomorrow by bridging critical gaps between hardware and software, while enabling the standardization to help our partners balance faster time-to-market and cost control alongside the ability to create their own unique solutions.”
Arm said Arm-based chips shipments have totaled more than 100 billion devices shipped over the last five years and claimed that at the current rate, 100 percent of the world’s shared data will soon be processed on Arm.
“Such pervasiveness conveys a responsibility on Arm to deliver more security and performance, along with other new features in Armv9,” the company said. “The new capabilities in Armv9 will accelerate the move from general-purpose to more specialized compute across every application as AI, the Internet of Things (IoT) and 5G gain momentum globally.”
On the security front, Arms said the Armv9 roadmap introduces the Arm Confidential Compute Architecture (CCA), which shields portions of code and data from access or modification while in-use, even from privileged software, by performing computation in a hardware-based secure environment.
“The Arm CCA will introduce the concept of dynamically created Realms, useable by all applications, in a region that is separate from both the secure and non-secure worlds,” Arm said. “For example, in business applications, Realms can protect commercially sensitive data and code from the rest of the system while it is in-use, at rest, and in transit.”
“The increasing complexity of use cases from edge to cloud cannot be addressed with a one-size-fits-all solution,” said Henry Sanders, corporate vice president and chief technology officer, Azure Edge and Platforms at Microsoft. “As a result, heterogeneous compute is becoming more ubiquitous, requiring greater synergy among hardware and software developers. A good example of this synergy between hardware and software are the ArmV9 confidential compute features which were developed in close collaboration with Microsoft. Arm is in a unique position to accelerate heterogeneous computing at the heart of an ecosystem, fostering open innovation on an architecture powering billions of devices.”
