Pulin Desai, Vision and AI Product Marketing Group Director at Cadence, presents the “Cadence Tensilica Edge AI Processor IP Solutions for Broad Market Use Cases” tutorial at the September 2020 Embedded Vision Summit.
In this talk, Desai presents the full range of Cadence Tensilica edge AI processing solutions. These silicon IP-based solutions serve markets from low-cost voice-activated consumer devices to high-throughput autonomous vehicle perception. They are widely deployed in high-volume AI-enabled end products such as smart speakers, mobile phones, surveillance cameras and automotive subsystems. Different applications have different requirements, and requirements evolve over time. In addition, diverse algorithms are often employed within an application for pre-processing, inference and post-processing.
Desai shows how the Cadence Tensilica edge AI processors meet the needs of specific markets, including distributing processing across low-power, programmable DSP and AI accelerators. He also highlights key trends in DNN topologies, and makes the case that, due to rapid progress in neural network research and varying processing requirements, programmable solutions are essential. Finally, he outlines the other elements of the Cadence Tensilica edge AI processing solutions, including software tools and support for a wide range of software frameworks.
See here for a PDF of the slides.