You may have already forgotten that the mobile device we hold in our hands today originated as a “cellular phone”, then became a “feature phone” and now a “smart phone”.
The features and functions we’re buying now and will buy in the next couple of years define something much more powerful and profoundly functional than a “smart phone”. New mobile use cases are emerging, demanding immersive and engaging user experiences, seamless scalability, and intelligent assistants to enrich and ease our lives. The “smartphone” is morphing into the gateway to the digital world. As technology is always changing, Arm is once again leading innovation for the mobile ecosystem.
This week, Arm rolled out a new suite of mobile solutions designed to tackle some of the biggest challenges the mobile design ecosystem faces in delivering solutions that bring these next generation experiences to life. Here’s what Arm believes are those challenges and how we can overcome them:
The lines between compute devices are blurring. Increasingly, desktop productivity experiences are delivered from a docked mobile device, enabling productivity wherever you are, with all data in one place – turning any place into a workplace. This puts some pressure on mobile designers who need, for example, to power 4K high dynamic range (HDR) displays while running multiple windows of apps simultaneously. Coupled with the fact that large-screen compute is becoming more “mobile-like” with, for example, apps on Chrome and Windows apps for the Arm instruction-set architecture, more single-thread performance and higher compute density in smaller form factors is more important than ever.
With these challenges in mind, the just-announced Arm Cortex-A75 CPU will improve the single-thread performance by 50% and the Arm Mali-G72 GPU will deliver 40% more graphical performance. The introduction early this year of Arm’s DynamIQ technology and last year’s introduction of the Mali Bifrost architecture will elevate performance and efficiency in these newer designs.
The unprecedented flexibility provided means that we are going to see a plethora of core combinations, such as 1x big + many LITTLE cores (e.g. 1b+3L or 1b+7L), Arm Cortex-A55 CPU. This is particularly exciting as it means that we will also see compute performance doubling in the more affordable smartphone price points. Nobody gets left behind!
The blurring lines in mobile design is just one part of the paradigm shift. There is also the ongoing virtual reality (VR) and mixed reality (MR) revolution to consider. While still in its infancy, we’ve seen the power and potential of VR/MR in specialty systems and use cases, but the biggest growth opportunity is by far in mobile. To drive those experiences into the mobile world, we need to overcome some hurdles, including:
All this computation also needs to be done within a fixed 2-3-watt thermal budget, pushing the boundaries on what’s possible with the ever-smaller devices.
While it sounds daunting, Arm engineers have had their eye on these potential barriers for a long time. The innovations in Bifrost and the scalability of Mali-G72 have been developed to address these key requirements and challenges to power the next generation of mobile VR devices.
With 40% more performance and 25% higher energy efficiency compared to the Mali-G71, Mali-G72 can support complex use cases for longer periods, without overheating your system or killing your battery. Scalability is also an ongoing consideration as we see partners choosing to implement more and more powerful configurations, which is why it was important to us to provide this 20% performance density improvement to enable more performance in a smaller silicon budget.
For a more in-depth look on how Arm is enabling engaging and immersive experiences, please read the “Enabling tomorrow’s technology today” blog.
The mobile experience as we know it is changing. It’s also clear that the new mobile paradigm is yielding more of a companion or personal assistant than just a smart phone. This new type of device requires powerful AI algorithms, some of which need to be computed on the device, rather than in the cloud, for offline availability, latency, privacy and security reasons. There are a number of initiatives currently aimed at bringing AI to mobile devices locally, such as TensorFlow Lite and Caffe2Go. As an example, think about facial and voice recognition, real-time “linguistic” responses and conversation, and more-accurate handwriting recognition.
Arm’s success in mobile is built on enabling applications, providing software platforms, and delivering the best-performing, most efficient underlying hardware through our semiconductor partnerships. That won’t change with AI. In fact, we are actively laying the foundation for future mobile with AI at the heart of those experiences. The performance delivered by Arm DynamIQ will enable increased use of AI across all devices, which we’ll see appearing in smartphone processors for 2018.
We’ve talked a lot about hardware, but hardware is nothing without software and that is why Arm is also working to ease the software development process for AI, ML and CV applications. We recently released the Compute Library, a collection of software functions optimized for Arm Cortex CPU and Arm Mali GPU architectures. This library will empower developers to get even more performance out of existing hardware, enable faster deployment and optimize functions for these new and emerging applications.
Compute solutions are changing rapidly and shaping our lives. The demand for intelligent performance is increasing across multiple segments to cater to new trends. With the launch of new Cortex-A CPUs based on DynamIQ technology, alongside Mali-G72—the highest performing Arm GPU—Arm and its partners continue to lay the foundations of tomorrow to support this new mobile computing paradigm.
We encourage you to visit the Arm DynamIQ: Technology for the next era of compute blog for a more in-depth analysis in how Arm is shaping the AI landscape.