Blog
Qualcomm’s Wear Elite diagram shows sensors at chip’s heart

Sensors continue to move far beyond commodity products to making possible intelligent edge devices and infrastructure.
In one new example, sensors are at the very heart of the new uber Snapdragon Wear Elite chip from Qualcomm, with support for more than 50 sensors in all, according to the company. That puts Wear Elite in step with, or possibly ahead of, other chip designers that are fusing sensing and sensing data with AI compute functions provided by GPUs, NPUs and CPUs.
So far, STMicroelectronics and Bosch Sensortec have stepped forward to say they support, respectively, an IMU for sensing motion and a barometric pressure sensor for detecting altitude on the new Qualcomm chip. However, Qualcomm hasn’t explained the full sensing potential for coming wearables, leaving that role presumably to OEMs it has named to use the chip in wearables: Google, Motorola and Samsung.
At a high level, the architectural drawing Qualcomm published for Wear Elite says volumes about the general role of sensing at the edge. Chip engineers take years to design new chips and want vital elements like sensing positioned in the proper place to reduce the length of interconnecting wires. Shorter distance means less time to travel to and from another element and those microseconds add up. An actual Wear Elite chip will have tiny wires connecting it to all the other elements, as do all the other elements, depicted as blocks in the drawing.
Notice how Sensing Hub, fits between GPU and Display blocks and also touches blocks for Micro-Power Connectivity and Fast Charge and Camera. Nearby are Video and Memory. Most consumers only care that a smartwatch or other wearable will work as advertised, but chip designers care about the chip functionality at the level of physics and how fast electrons move through a system. (It helps it doesn’t cost too much to make!)
Given the Qualcomm drawing is likely not a true representation of the actual chip, it is still important that the Sensor block sits close to compute and memory functions, noted Jack Gold, president and principal analyst at J. Gold Associates. He explained to Fierce: “Many sensors are real time, but many are not. So for real time sensors where you need to measure and react quickly, having a short path to compute and memory are important. Perhaps not so much for sensors that only take a reading every few seconds, minutes and hours. For the most part, compute cycles in modern chips are short enough that a reaction to sensor data is almost instantaneous no matter where in the silicon the sensor interface is placed. That’s also true for AI capable systems. The bigger issue, in my opinion, is how to manage the power needed for sensor interconnect when you are talking about a really small system that Wear is targeted at.”
Okay, Gold is right. The Qualcomm drawing may be more figurative than literal, but the centrality of sensors remains undeniable. As he noted, the power management connection for a wearable like a ring or pendant might be the biggest management concern. Besides, physical position on a board will not determine all the elements of power management. Software will manage power flows and software design functions remain well within the domain of OEMs who design entire systems.










