The Linley Group - Linley Fall Processor Conference 2019

Neural networks are graph-based applications with opportunities to execute many graph nodes concurrently. Recent architectures have responded with massively parallel systems, but scheduling them has proved challenging, often relying on an oracle compiler. Instead, Mythic created an architecture that works on graphs directly: producer/consumer relationships are hardware concepts, and parallel execution happens automagically when dependencies are met. This presentation discloses a high-level overview of the architecture and how it can efficiently achieve parallelism on a wide variety of neural networks.