DEEPX Demonstrator

• The demo shows 16 camera streams processed simultaneously using one single DEEPX DX‑M1 AI accelerator
• All AI inference is handled by the DX‑M1 chip, not by the CPU or GPU
• The camera streams are processed in parallel, enabling real‑time performance with no visible latency (“zero lag”)
• The system runs on a standard AMD Ryzen Embedded 8000 series platform under Ubuntu Linux
• The DX-M1 is integrated as an M.2 accelerator module, making system integration simple
• This demonstrates high multi-stream scalability with low system complexity
• Using one AI chip instead of multiple accelerators reduces:
  • System cost (BOM)
  • Power consumption
  • Thermal and design complexity
• The setup is available today, meaning it is not a roadmap concept but a deployable solution

Exhibitor: Avnet Silica