Wear OS smartwatches have a dual-chipset architecture inclusive of a powerful application processor (AP) and ultra low-power co-processor microcontroller unit (MCU). The architecture has a powerful AP capable of handling complex operations en-masse, and is seamlessly coupled with a low power MCU.
The Wear OS hybrid interface enables intelligent switching between the MCU or the AP, allowing the AP to be suspended when not needed to preserve battery life. It helps, for instance, achieve more power-efficient experiences, like sensor data processing on the MCU while the AP is asleep. At the same time, the hybrid interface provides a seamless transition between these states, keeping a rich and premium user experience without jarring transitions between power modes.
Kseniia Shumelchyk on the Android Developers Blog
The new OnePlus Watch 2 is the first to use this new architecture, and the most interesting part is that it runs not one, but two operating systems: Wear OS, which is Android, running on the “AP”, and a smaller RTOS that runs on the “MCU”. In the case of the OnePlus Watch 2, the “AP” is a Snapdragon W5, while the “MCU” is a BES 2700, an ultra low power microcontroller.
I can’t seem to find any information on this “RTOS”, but I’d really love to know what it’s based on.
That’s not a surprise though. Snapdragon w5+, which is the w5 version which should be used in non-penny pinching designs, also contains an always-on processor. And that’s a processor by Qualcomm, rather than an obscure producer quite probably affiliated with the Chinese government.
The BES 2700 from Bestechnic is basically a bluetooth controller than can also drive the display. It uses an ARM Cortex M (like Micro Controller) core which is a pretty generic solution. AFAIK there are already some SoCs based on ARM Cortex A (like Application Processor) that also incorporate the Cortex M on die.
There are plenty of RTOS for Cortex M around, with FreeRTOS being a very common one. Typically usefull applications can already be built on 32kB RAM.
Which begs the question: Why can’t RTOS run the smartwatch all by itself? What is in Android Wear that makes it necessary on a device like a smartwatch (not a smartphone or tablet)?
Anyone else thinks we lost big time when Pebble went under?
There are smartwatches that don’t use Android Wear, like my Huawei Watch GT2. SO I have 14 days on battery, but lack of Android Wear app ecosystem. I’m not developer, but I guess it’s easier to make apps and Integra te them with Android by using Android Wear platform. However i choose long standing. Charging watch every 2-3 days doesn’t sounds good to me.
Agreed, I have the T-Rex 2 by Amazfit and I get over a week solid with the always-on display setting, I could probably get up to three weeks with the screen set to only come on when in use. It has a rich app ecosystem, geared mostly towards health and fitness, but I got it for the battery life, price, and looks. It also has sensors that my old Apple Watch SE at twice the price didn’t have, like EKG and SpO2.
I certainly don’t miss a smartphone-class CPU and the corresponding battery life measured in hours instead of weeks.
I think the so~nution would be take this approach to one step beyond: to abandon Linux for areal time microkernel with “basic” task dealt by the microcontroller and “complex” task dealt by the more powerful CPU.
What we are seeing here has its limitations.
Like Marshal says, the eco system of dev tools is nicer with Android wear than something like the pinetime options https://wiki.pine64.org/wiki/PineTime_Development
Pebble was very nice to use, I only used it after it was basically already dead so I didn’t invest in learning the SDK.
Having gone through many smart watches, this is not always an easy question.
If the “watch” has only minimal functionality, and is expected to work tethered to a phone, yes it can be done. I used to have a Pebble (yes!), which might have represented the ultimate device in this category (eInk screen, very long runtime).
However if you want more than basic functionality, like:
– Always on (non eInk) screen — no need to wiggle your wrist to actually see the time
– WiFi connectivity, or more than Bluetooth LE
– GPS for activity tracking
– If you are fancy GSM modem for complete standalone use
– and of course multimedia apps
There is not much choice.
However I really like this idea. “Dual” OS devices have been done in the past (or more common than we think, as all phones have at least two operating systems, the other one being “baseband”). And we have ultra low power cores.
So, maybe the hybrid will save us (next question is: can we have an OLED/eInk hybrid as well)?