Can ARM chips compete neck-and-neck with Intel and AMD on benchmarks? That could be happening sooner than you think.
Starting next year, ARM processors will get significantly faster thanks to big changes in the company’s Cortex-A chip designs. ARM is taking a page from rivals like AMD that have focused on raising the performance threshold in chips.
ARM isn’t known for superfast chips; it is instead mainly associated with power-efficient chips that give long battery life to devices. That focus has helped the company succeed in mobile devices, an area where Intel’s power hungry chips failed.
But applications like virtual reality and machine learning need more performance, and ARM is preparing its processors to take on those emerging applications. ARM is adding more cores, instructions, and faster pipelines in smaller spaces to boost performance.
The new features are packaged under the term “DynamIQ” and will be in upcoming Cortex chips for mobile, PC, server, and internet-of-things devices.
Performance metrics will come when ARM announces the first chip designs with DynamIQ features, which could happen early next year, said John Ronco, vice president of product marketing.
DynamIQ adds more performance without compromising ARM’s power efficiency focus, Ronco said. Most devices from ARM don’t require cooling fans, and that will remain the case with DynamIQ features in chip designs.
The improvements increase the possibility of Apple switching over to ARM architecture on its Macs. That hasn’t happened yet as Intel’s mainstream Core chips are significantly faster than current ARM processors. Like other chipmakers, Apple licenses ARM architecture, which it already uses in iPhones, iPads, and other devices.
It’s becoming a trend for chipmakers to pile on more performance in chips, especially with the emergence of gaming and virtual reality. Intel is committed to delivering a minimum of 15 percent more performance with every new chip generation, and AMD said its new Ryzen chips delivered 40 percent more instructions per cycle (IPC), an important performance metric.
Architectures get modified with market demands of the time, and the chip industry is seeing the emergence of machine learning as a primary workload, said Dean McCarron, principal analyst at Mercury Research.
Though ARM is a RISC design, it is getting new extensions and evolving much like x86, making the two hard to differentiate, McCarron said.
ARM previously added vector extensions to its new ARMv8-A SVE (Scalable Vector Extensions) design for high-performance computing. The company has also added extensions for virtualization and other features.
The new DynamIQ features will be in the existing ARMv8-A architecture. Cortex designs will be able to pack up to eight cores in a single cluster, compared to four cores in existing designs. The cores will share memory, throughput, and other core components.
A chip could have different types of cores, for example, an eight-core chip with four fast CPUs for demanding applications and four slower CPUs for low-power, fast-moving tasks. That is already a part of ARM’s “Big.Little” design but is being enhanced with DynamIQ.
MediaTek’s X30 chip already has 10 CPU cores concentrated in three packages. In theory, DynamIQ will allow Cortex-based chips to pack 16 cores in two clusters, making better use of limited smartphone space.
Also in the DynamIQ are new instructions for machine learning, and CPUs will deliver 50 times more artificial intelligence performance in the next three to five years, ARM claimed.
At the heart of these improvements are the ability for CPUs to do more low-level and power-efficient processing of machine-learning tasks. All of the calculations can be combined to provide an estimated answer to a question. The improvements include more half-precision floating point operations, which are common in machine-learning-focused chips like Intel’s upcoming Knights Mill, and AI GPUs from AMD and Nvidia.
A big performance boost will also come from faster pipelines, but there is no fundamental change in the core pipeline design or branch prediction. Radical changes there are more of a long-term project, Ronco said.
ARM has older processors like the Cortex-A8, which is more than a decade old and recently used in the BeagleBone Blue robotics developer board. It won’t be possible to mix those CPU architectures into the new DynamIQ design.