Mapper/Router Methodology - 2024.1 English

AI Engine-ML Kernel and Graph Programming Guide (UG1603)

Document ID
UG1603
Release Date
2024-06-06
Version
2024.1 English

For more information on mapper and router methodology, see Mapper/Router Methodology in the AI Engine Tools and Flows User Guide (UG1076).

In general, the methodology for mapping/placing the ADF graph onto the AI Engine-ML array remains similar to AI Engine-ML 1. You should keep the updated AI Engine-ML 2 architecture in mind (new AXI memory access patterns, presence of additional memory tile row) while constraining the design. The major change is that AI Engine-ML 2 does not have DMA FIFOs, and you cannot specify DMA FIFO as a type for fifo_depth() constraints. In addition, you can now specify location constraints on shared buffers, and external buffer ports.