Arria® 10 or Cyclone® 10 GX Avalon® Memory-Mapped (Avalon-MM) DMA Interface for PCI Express* Solutions User Guide

ID 683425
Date 9/10/2024
Public
Document Table of Contents

10. Arria® 10 or Cyclone® 10 GX Avalon-MM DMA for PCI Express

The Arria® 10 or Cyclone® 10 GX Avalon-MM DMA for PCI Express IP Core includes highly efficient Read DMA, Write DMA, and DMA Descriptor Controller modules. The typical throughput for hardware systems using a 256-bit interface to the Application Layer is 6 GB/sec or higher. For hardware systems using a 128-bit interface to the Application Layer the throughput scales proportionately to 3 GB/sec.

Using a 64-byte payload, the maximum theoretical throughput is far less due to the increased proportion of the bandwidth taken by the TLP headers. The throughput for back-to-back TX memory write completions, RX read completions, and simultaneous reads and writes is 2 GB/sec.

Note: A 64-byte packet is the minimum packet size for Ethernet.