Launched Arm, the British tech company whose chip technology runs among most smartphones in the world, is a new generation of technology aiming to acquire data centers and AI applications ahead of its planned $ 40 billion merger with Nvidia.
Arm has released a new version of its so-called instruction set architecture, which specifies the type of software that can be run, and the company performs major updates to the architecture only once per decade.
Intel is its main competitor in this area, offering the so-called x86 architecture that powers most laptops and data center servers today based on chips from Intel and AMD.
The new architecture, called Armv9, aims to confront Intel in three main areas: security, artificial intelligence, and overall computing power.
One of the major developments for Intel and AMD chips has been the ability of cloud computing providers to use software to split the chips into virtual machines that can run applications from multiple clients, which in turn helped companies such as Amazon Web Services and Microsoft Azure generate billions of dollars in profits.
Arm introduced a concept called Realms, which accomplishes a similar task in chip hardware, and Simon Segars, CEO of the company, said during a press release: We see no reason why Realm technology cannot expand to provide security in the most high-performance systems out there.
Arm also said: It is adding features to AI chips for tasks such as image and speech recognition, an area in which Nvidia chips have outperformed Intel and helped it outpace Intel by market value.
“We are excited about the possibility of integration with Nvidia and the enhanced engineering capabilities that we will be able to combine with the ubiquitous computing platform with its deep expertise in the field of artificial intelligence,” added Segars.
The ARMv9 architecture includes the company’s new trusted computing architecture, or CCA, that enables the new Realms concept, and the company says: All applications have the opportunity to take advantage of Realms to protect code and any private data it processes through implementation in an environment different from other programs.
ARMv9 chips also contain the latest version of the company’s SVE technology, which was developed with Fujitsu’s help, and is one of the technologies used within the world’s fastest supercomputer.
The company says: The new SVE2 platform gives ARMv9 chips an edge when processing 5G, machine learning, and virtual and augmented reality workloads locally via the CPU.
And when it comes to computing power, ARMv9 enables a performance increase of more than 30 percent over future generations of mobile and cloud CPUs, thanks in part to the Total Compute design philosophy.
ARM did not say when the first ARMv9 chips would reach consumers, but the company is optimistic about its prospects.
ARMv9 powers 300 billion new ARM-based chips, and the company estimates that 100 percent of all shared data in the world will soon be processed by ARM chips, be it on a single device or in the cloud, it says.