Single-Thread Linear Read - 1.0 English

Versal Adaptive SoC Programmable Network on Chip and Integrated Memory Controller 1.0 LogiCORE IP Product Guide (PG313)

Document ID
PG313
Release Date
2023-11-01
Version
1.0 English

A single device accesses the entire memory linearly. Access within a page (row) is very efficient, but switching pages can cause loss of efficiency. To minimize the loss, choose a mapping that will switch banks at page boundaries and, furthermore, alternate between different bank groups after each 64-byte DRAM command.

The chosen mapping is therefore: 16R-2B-1BG-7C-1BG-3C.

This mapping places a bank group bit after three column bits. This ensures that back-to-back DRAM commands target different bank groups and avoid same-group penalty. For comparison, the following table shows performance with other possible mappings.

Table 1. Single-Thread Linear Read Address Mapping
Mapping Efficiency Mapping Type
16R-2B-1BG-7C-1BG-3C 95% Row-bank-column with bank group optimization
16R-2B-2BG-10C 54% Simple row-bank-column