The non-linear increase in demand in machine learning and other compute intensive applications leads to the development of the AMD Versal™ adaptive SoC AI Engine-Machine Learning (AIE-ML). The AIE-ML, the dual-core Arm® Cortex®-A72 and Cortex-R5F processor (PS), and the next generation programmable logic (PL) are all tied together with a high-bandwidth NoC to form a new architecture in adaptive SoC. The AIE-ML and PL are intended to complement each other to handle functions that match their strengths. With the custom memory hierarchy, multi-cast stream capability on AI interconnect and AI-optimized vector instructions support, the Versal adaptive SoC AIE-MLs are optimized for various compute-intensive applications, for example machine learning inference acceleration in data center applications by enabling deterministic latency and low neural network latency with high performance per watt.