TinyML Summit 2021 Tiny Talks - TinyML Software Runtime for Hybrid Multicore Architecture

Explore the intricacies of TinyML software runtime for hybrid multicore architecture in this insightful 24-minute talk from the tinyML Summit 2021. Delve into the challenges of running optimized neural networks on complex embedded hardware, focusing on hybrid multicore systems that combine CPU, DSP, and NPU cores. Learn about the importance of efficient resource allocation, minimizing processing overhead, and reducing memory transfers. Discover how Eta Compute addresses these challenges with their TENSAI Flow runtime and executors, and gain valuable insights into the industry's approaches to optimizing TinyML performance on edge AI hardware.

Introduction
Agenda
Requirements
Dependency
Data Compute
Compiler
Sponsors