Deploying AI Models at the Edge

Qualcomm Technologies demonstrates generative AI on Android phone

Qualcomm Technologies demonstrates generative AI on Android phone

Qualcomm Technologies, Inc. has announced a breakthrough in artificial intelligence (AI) with the first on-device demonstration of Stable Diffusion, a popular foundation model, on an Android phone. This is a significant milestone for Qualcomm AI Research as it proves that AI models can be run efficiently on edge devices without any reliance on the cloud or an internet connection.

Stable Diffusion is a generative AI model from Qualcomm AI Research that can generate pictures based on text prompts within ten seconds. For example, it can take the words “sunset” and create an image of a beautiful sunset with colors and shapes that look almost photorealistic. This model also has image editing, in-painting, style transfer and super-resolution applications.

Qualcomm says the on-device demonstration of Stable Diffusion was made possible with the advances in full-stack research and optimizations to run this model efficiently on phones.

“Our full-stack AI research means optimizing across the application, the neural network model, the algorithms, the software, and the hardware, as well as working across disciplines within the company,” states Qualcomm in the company’s OnQ blog.

Qualcomm says this is compelling because on-device processing with edge AI provides reliability, latency, privacy and cost-efficiency benefits.

Qualcomm Technologies’ optimizations for Stable Diffusion allow it to be run efficiently on phones and other platforms, preserving user privacy while providing low-cost edge AI processing.

“The result of this full-stack optimization is running Stable Diffusion on a smartphone under 15 seconds for 20 inference steps to generate a 512×512 pixel image — this is the fastest inference on a smartphone and comparable to cloud latency,” explains Qualcomm.

This ensures faster time-to-market for the next foundation model, allowing edge AI to become more ubiquitous across devices.

For example, Qualcomm’s one AI technology roadmap allows efficient scaling across different devices and models, such as laptops and XR headsets.

“Our one technology roadmap allows us to scale and utilize a single AI stack that works across not only different end devices but also different models,” states Qualcomm.

These optimizations mean that the time-to-market for the next foundation model the company wants to run on the edge will also decrease.

“Our vision of the connected, intelligent edge is happening before our eyes as large AI cloud models begin gravitating toward running on edge devices, faster and faster,” adds Qualcomm. “What was considered impossible only a few years ago is now possible.”

Article Topics

 |   |   |   |   |   | 

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Featured Edge Computing Company

Edge Ecosystem Videos

Automating the Edge

“Automating

Deploying AI Models at the Edge

“Deploying

Latest News