Ready to enter a new era of AI-ready smartphones?
- The Qualcomm Snapdragon 888 chip will let handsets take high-resolution multimedia faster and perform AI-related tasks more efficiently, plus 5G connectivity
The unveiling of Qualcomm’s new flagship chip, the Snapdragon 888, for Android smartphones underlines just how big a role artificial intelligence (AI) capabilities will play in upcoming smartphones.
Just like how the recent iPhone 12 launch unpacked the tiny transistors and the AI-cum-TinyML advantages of Apple’s new A14 Bionic chip – promising nearly 30% uptick in CPU, GPU, and AI-powered sensors performances – the Qualcomm Snapdragon 888 is the chipmaker’s first 5-nanometer chip.
The smaller transistors on these 5nm chips mean that performance gains are boosted up to 35% higher than previous generations, and the 888 also has its 5G modem built into it rather than a separate component in previous chips – which means that connectivity is also being enhanced alongside the processing and graphical capabilities.
The combination of 5G coverage and AI-powered sensor input should be no slouch for the next-generation of topline Android devices, but AI will not just be enhancing the camera and other sensors anymore. “Qualcomm’s addition of a large AI accelerator and TinyML to its next-generation 5G mobile platform is a game-changer for 5G mobile devices,” said Lian Jye Su, the principal analyst at tech market advisory firm, ABI Research.
“As more users switch to 5G mobile devices in 2021, embedded AI will be a critical success factor for enhanced user experience.”
Qualcomm says this Snapdragon features a completely re-engineered AI Engine, the Hexagon 780. The chipmaker believes this redesign will help software developers working on Snapdragon 888-powered devices to make improvements in a number of areas including live-motion tracking of objects (to assist autofocus in video and photography, for instance); automatically adjusting audio to account for the device’s surrounding ambient environment; improving augmented reality filters in AR-capable apps.
Qualcomm’s demos of the Hexagon 780 in real-world use-cases show just how well-rounded the AI processing is in a multitude of functions including in better natural language processing, translation, even cutting down the battery usage, with a performance-per-watt output that is three times higher than the Snapdragon 865. The AI computational performance is said to be so remarkable that Qualcomm says that the Snapdragon 888’s AI processing can erase a recorded character from a video scene and insert someone else instead – in real-time.
“With the AI capabilities of Snapdragon 888, OEMs can pack more and more AI enhancements into various applications in their mobile device, including beautification, edge detection, depth sensing and low-light noise reduction in the camera, more secure facial and fingerprint unlock in biometric identification, more fluid gaming performance and smarter app recommendations and battery optimization,” noted ABI’s Lian. “Qualcomm can achieve this by allowing the AI engine to access a large pool of shared memory to enable mobile devices to carry larger AI models that are more powerful and accurate.”
The introduction of a dedicated TinyML chipset will also bring ambient sensing capabilities to the next generation of smartphones, such as noise reduction, Wi-Fi, and Bluetooth signal optimization, not to mention ‘always-on’ machine vision. The breadth of TinyML applications and sensing capabilities is indicative of its untapped market potential, with ABI Research forecasting that a total of 2.5 billion devices will be shipped over the next ten years with TinyML chipsets.
And while the Qualcomm Snapdragon 888 will be the new flagship jewel for Qualcomm-enabled Android smartphones, the best part is that the wide-spanning AI capabilities of the premium devices should make their way to other smartphone price tiers as well, not just reserved as performance benefits for the topline mobile devices.
As ABI Research’s analyst Lian concluded: “In the long run, expect all these AI capabilities to trickle down into Qualcomm’s mid-tier and low tier chipsets, bringing democratization of AI across all types of mobile devices.”