Latency and Resource Trade-off Analysis of AXI-DMA and BRAM Integration Approaches on SoC-FPGA

Feb 28, 2026 by G. Tatar (Firat University Journal of Experimental and Computational Engineering)

DOI 10.62520/fujece.1790038

We compared AXI-DMA vs BRAM integration for a custom PE on Zynq-7000 and showed a clear scalability vs latency trade-off: AXI-DMA buys streaming flexibility and throughput at ~54% higher resource cost, while a BRAM-based design slashes latency for small-scale ops (about 0.10 ms faster per operation) with much lower area—practical guidance for picking integration style when building FPGA accelerators on heterogeneous SoC platforms.

source S2, crossref