Talk / Overview

New model architectures, model quantization, optimization techniques and innovative hardware design bring the power of AI and ML to very low cost processors. Synaptics low cost and high performance audio and video Edge SoC (Systems-on-a Chip) are now enabled with hardware acceleration for AI. This talk provides a short history and status on some AI inference engines, then presents and explains some of the software and hardware techniques developed and used by Synaptics to accelerate models and finally provides some comparative benchmarks on popular computer vision models running on Synaptics SoCs.

Talk / Speakers

Abdel Younes

Technical Director – Software Architecture, AI/ML Technologies, IoT Division, Synaptics

Gaurav Arora

Vice President - System Architecture and AI/ML Technologies, IoT Division, Synaptics

Talk / Highlights

19:34

Model quantization and Hardware acceleration, how fast can we get?

With Abdel Younes & Gaurav AroraPublished March 12, 2020

AMLD / Global partners