AI Engine is formed from several distinct technologies which combine to enable high-performance AI. The ecosystem is targeted at user-facing AI experiences on next-generation mobile platforms. Apps supported by AI Engine will run all their processing locally, so they have improved responsiveness and work offline.
AI Engine runs atop Qualcomm’s processor architectures including its Kyro CPU, Adreno GPU and Hexagon Vector Processor. These chips are engineered to support performant artificial intelligence with minimal overheads. AI Engine then augments the hardware capabilities with several software layers that expose the chips to app developers.
Technologies available through the AI Engine include Qualcomm’s Snapdragon Neural Processing Engine (NPE) software framework, the Android Neural Networks API and the Qualcomm Hexagon Neural Network library. While the framework used will vary depending on the developer’s aims, these subsystems permit direct access to the device’s hardware. This increases flexibility and eliminates the overheads traditionally associated with on-device AI processing.
READ NEXT: AI cybersecurity provider Vectra raises $36m in funding
Adoption of on-device AI is growing in apps. Social media products such as Snapchat and Facebook use AI to power new masks and camera effects. The technology’s also being used for advanced biometric authentication, computer vision and cybersecurity solutions. Qualcomm’s AI Engine is intended to accelerate the performance of these apps, allowing for more demanding on-device AI.
“On-device AI demands multiple hardware architectures, software tools, and frameworks to meet developer demand and performance thresholds for the myriad of AI-powered features and apps arriving on smartphones this year,” said Gary Brotman, Qualcomm director of product management.
“The AI Engine capabilities of select Snapdragon mobile platforms have and will continue to fuel AI innovation among our OEMs and ecosystem partners, and are engineered to accelerate the development and delivery of rich, seamless and compelling features for consumers.”
Qualcomm provided several examples of apps that already use Snapdragon platform optimisations to improve their AI. Tencent’s popular Mobile QQ social game is using AI Engine to improve frame rates in some of its new features. Baidu is also preparing to deploy the technology. Support is growing from phone makers too, with Xiaomi, OnePlus, OPPO, Motorola, ZTE and others all planning to add AI Engine to devices.
Qualcomm will be demonstrating the technology at its booth at Mobile World Congress next week. AI Engine will be compatible with Snapdragon mobile platforms including the 845, 835, 820 and 660 chips. Qualcomm claims best performance will be obtained with its upcoming Snapdragon 845 flagship.
