Explore the world of ultra-low power machine learning systems through this 24-minute conference talk from the tinyML Summit 2020. Delve into the concept of tinyMLPerf, a benchmarking tool for tiny machine learning systems, as presented by Vijay Janapa Reddi, MLPerf Inference Chair and Associate Professor at Harvard University. Gain insights into the methodology, ecosystem impact, and unique aspects of tinyML, including quantization and pruning techniques. Understand the community-driven effort behind tinyMLPerf, its set of tasks, and how it enables performance evaluation in the rapidly evolving field of ultra-low power machine learning.