Qualcomm Ventures Beyond Mobile Processor Market in New Roadmap
The new Snapdragon extension expands the mobile platform's capabilities to new markets from PCs to ADAS.
Qualcomm's Snapdragon platform is often associated with its intelligent mobile processor. Now, the company is pushing to expand connectivity beyond smartphones. At Qualcomm Investor Day, the company’s CTO Dr. James Thompson presented the new roadmap, explaining an ecosystem “where everyone and everything is connected.”
Target areas of expansion in Qualcomm's One Tech Roadmap. Screenshot used courtesy of Qualcomm
Qualcomm aptly named its new directional plan the "One Tech Roadmap for the Connected Technology Edge," indicating a significant expansion from mobile to all devices across the board.
From PCs to Connected Vehicles
The new Connected Tech Platform goes beyond Android to support different operating systems. It incorporates advanced automotive cameras with gaming-class graphics and computer vision capable of recognizing a pedestrian quickly. Graphic quality will be brought from smartphones to PCs and used for extended reality (XR).
PCs will be upgraded with a next-generation CPU, including fully-integrated CPU and GPU accelerators. This CPU is said to enable intensive processing power for AI operations at the edge.
Data coming from every sensor in a connected automotive system devours power; every car processes approximately 5 Gbps to 15 Gbps, as shown in BMW's ADAS. With Qualcomm's expanded processing hardware, however, a portion of the data will now be processed on the edge instead of relying on the cloud completely.
Such hardware complexity requires complex software. Qualcomm will offer the Neural Processing SDK for AI at the core of scalable hardware and frameworks.
The development workflow for Qualcomm's Neural Processing SDK. Image used courtesy of Qualcomm
Other Highlights: Camera, Graphics, Processing, and Connectivity
Qualcomm’s smartphone camera offers performance that has appealed to smartphone giants like Xiaomi. Now, the company intends to intersect AI/ML algorithms with camera functionality; the camera will identify the subject of a photo and improve the quality with semantic segmentation.
New Qualcomm high-dynamic-range cameras are described as having a better signal-to-noise ratio. Computer vision will offer dense optical flow, tracking motion for every pixel. In automotive systems using the Snapdragon Ride hardware platform, the camera could use this computer vision to immediately identify moving pedestrians. In XR, this technology can be paired with eye-tracking to create a clearer picture of the environment and reduce graphics processing as a result.
Qualcomm also plans to scale graphics for PCs—for instance, for gaming—along with automotive and other connected systems.
The company's roadmap may accelerate signal processing-based technologies with AI to make computation more energy efficient. Qualcomm's processing advancements have been expedited by its acquisition of Nuvia, a supplier of Arm CPUs that may extend Qualcomm's processors to Windows and Chrome.
The roadmap reveals how Qualcomm will track hardware advances alongside 5G releases. For instance, Release 17 of 5G is expected in two years; at that time, Qualcomm will welcome new features and enhanced brand aggregation into its smartphone chips to cover higher millimeter waves. Beyond that, 5G Release 18 will include new types of devices (like vehicle-to-vehicle and vehicle-to-pedestrian connection devices) and create a fertile ground for AR. The expected network increase capacity is estimated at 50 percent.
Qualcomm's power-amplifier design for modem-to-antenna performance. Screenshot used courtesy of Qualcomm
Qualcomm's Goals for the Roadmap
The advantages of Qualcomm's roadmap are numerous.
First, it supports worldwide customers with an existing infrastructure that has 10,000 different band combinations. Second, it can support large, complex architectures because it controls the system from the baseband to the antenna. Third, the addition of ultraBAW (bulk acoustic wave) filters will differentiate frequencies from 600 MHz to 7.2 GHz. Together with ultraSAW filter technologies, ultraBAW can cover a long range of the radio spectrum.
Qualcomm offers transistors, digital pre-distortion (DPD) for power amplifiers, and envelope processing—all of which enable modern RF front-end modules (among other components). These modules can support eight digital streams and different devices.
An overview of Qualcomm's modem RF offerings. Screenshot used courtesy of Qualcomm
Finally, this roadmap supports the vision for a future free of wires and cables.