-
Notifications
You must be signed in to change notification settings - Fork 20
OSNT DMA (deprecated)
BE AWARE this page is valid ONLY for OSNT releases 1.0.0 and 1.5.0
osnt_dma
v1.00a
Yilong Geng (gengyl08_at_stanford.edu)
pcore (HW)
lib/hw/osnt/pcores/osnt_dma_v1_00_a/
AXI4-Stream
AXI4-Lite
PCIe
This module is built upon the original NetFPGA DMA. The new DMA copies the received packets into a ring buffer in the host directly, which is useful for high speed applications. There is one RX completion ring buffer (4KB) and one packet ring buffer(64KB). For each packet received, there will be a 64B RX completion message in the RX completion ring buffer and the actual packet is copied in packet ring buffer. When packets are small, the RX completion ring buffer will be full first with 64 packets. When packets are large, for example of one MTU size, the packet buffer will become full first with 42 packets. So the number of packets the ring buffers can hold depends on the size of the packets, varies from 42 to 64.