At ARM, we’ve had the opportunity to work with our partners to define the mobile revolution and now we’re hard at work doing the same for the growing wearable devices category. We’re simply cracking the surface of wearable technology – a market set to expand to $30 billion a year by 2018 according to market research firm IHS.
And though we’re in early days, we’re already coming to some important conclusions on how best to build a wearable device that’s efficient, accurate, powerful and useful. At its core, the key challenge in wearable devices is achieving an all-important balance between processing power and energy efficiency – and it’s at this place that ARM’s expertise and pedigree has helped us lead the way in architecting wearable technology.
A deep dive into our position on wearable technology can be found in a new ARM whitepaper but let’s look at a few of our findings on how we can help build the best wearable devices.
One of the most interesting and challenging parts of the wearable technology sector is addressing just how diverse they can be – encompassing simple fitness trackers to high end smart watches and everything in between. They need to be able to take in and process data streams from any number of sensors including accelerometers, gyroscopes, GPS antennas, temperature and pressure sensors. They can run anything from a simple real time OS to a rich OS implementation such as Linux or Android. And in this diverse landscape there’s a common set of challenges, the need to be “always-on, always-aware” while also being lightweight and highly power efficient.
ARM has been a major innovator in this space and products like the Cortex-M are especially designed to maximize power efficiency. These chips are perfect for fitness trackers or simple smartwatches – providing just enough processing power and incredible power efficiency. And if you need a bit more power, the Cortex-A series can enable advanced performance without adding a large drain on energy. Coupled with a Mali GPU, these chips can bring a rich graphical user experience to a smartwatch or similar device.
Wearables bring a whole new power class requirement vs. that of a traditional smartphone. For example, the average smartphone requires a daily charge of around 3,000 mAh, but a wearable device needs to use around 300 mAh per week… For wearables to make sense for end users, they need to be able to work often and charged infrequently. But even so, we can use the knowledge gained in our experience with smartphones, tablets and other mobile devices as a jumping off point. At ARM, we’ve been working on optimizing SoCs for use in wearable devices. This includes things like using smaller memory caches to save on both die area and power consumption – making our chips smaller and more energy efficient. And these changes are meaningful: we’ve been able to halve the L1 cache size from 32K to 16K while only experiencing a 10 percent impact on performance.
One of ARM’s greatest strengths is our vibrant and robust partner ecosystem. It means that we have the privilege of tackling the challenges of wearable devices from all angles and across the diverse spectrum of wearable technology. We fully expect to see this category of devices blossom in the coming months and years – and with it, we’ll discover new and better ways to address the specific needs of wearables. We look forward to seeing where this path takes us and fully expect to see uses and form factors that we’ve not yet even imagined. And throughout the journey, we’re confident that ARM offers the best platform – providing an industry leading balance of processing power and power efficiency – on which to build the next generation of wearable devices.
Like! Especially good how the whitepaper defines architectures and positions cores and interconnect in a very popular and hot wearables context.This, combined with giving good insight into our partners' relevant offerings and providing power/performance/size data makes this a very good info source for anybody working with wearable high-tech!
Great to see the Whitepaper published! Think this might also be of interest to followers of the ARM Processors group and the News space