Radio-ML ConvNet: Throughput - 2025.2 English - XD100

Vitis Tutorials: AI Engine Development (XD100)

Document ID
XD100
Release Date
2025-12-05
Version
2025.2 English

From the following figure, you can calculate what the throughput is based on AI Engine simulations. Eight inferences are ran in 532.8us, which translate to ~15,000 inferences per second.

figure