Deploy and scale at the embedded edge.
12x faster end-to-end pipeline FPS than PCIe ML accelerator.
Significant advantage over competition in both performance and power.