MediaTek Unveils On-Device GenAI Demos at MWC 2024

MediaTek, a leading global fabless semiconductor powering more than two billion connected edge devices every year, is showcased a series of new Generative AI demonstrations at Mobile World Congress 2024, all of which are utilizing the latest generation of APU hardware featured in the Dimensity 9300 flagship smartphone SoC. These live exhibitions will highlight several key Generative AI technologies enabled by MediaTek, including multiple industry-first applications running entirely on-device.

MediaTek AI Enabler

“As an enabler of AI, we want these industry-first experiences to inspire developers, OEMs and partners through MediaTek-powered AI technology, such as the APUs integrated into the Dimensity 9300 and 8300,” said JC Hsu, Corporate Senior Vice President at MediaTek. “We are seeing the tip of the iceberg on the Generative AI use cases that can be created, and we’re looking forward to working closely with others to turn visions into reality.”

Attendees at Mobile World Congress had the opportunity to experience the following hands-on demonstrations of key on-device AI technologies, each of which are industry-firsts:

• SDXL Turbo: A text-to-image Stable Diffusion engine that dynamically generates images in real-time, based on user-provided prompts.
• Video Diffusion Generation: Enables fast video diffusion generation, capable of a range of animation styles.
• Low Rank Adaptation (LoRA) Fusion: In this demo, MediaTek is showing how it has incorporated Low Rank Adaptation (LORA) Fusion into the NeuroPilot AI platform, exhibited in a use case in which animated avatars are created from a camera and processed in real-time.

These demonstrations were showcased at Mobile World Congress February 26-29, at the MediaTek Booth, Hall 3 #3D10.