Page History

...

• Support "point-to-point" and "network" mode
  • "point-to-point" by default (better logic optimization)

• Layer 2 Ethernet framing

  • Wrapper on the Ethernet MAC
  • Targeting 100G Hard IP MAC
    
    • 512-bit AXI Stream interface
• Chuck up the stream to MAX_SIZE=1024B (default)
  
  • 92.1% efficient = 1024B/(1024B + 4B FCS + 12B intergap + 8B preamble + 64B PGP Ethernet Header)
  • 95.9% efficient = 2048B/(2048B + 4B FCS + 12B intergap + 8B preamble + 64B PGP Ethernet Header)
  • 97.9% efficient = 4096B/(4096B + 4B FCS + 12B intergap + 8B preamble + 64B PGP Ethernet Header)
  • 98.9% efficient = 8196B/(8196B + 4B FCS + 12B intergap + 8B preamble + 64B PGP Ethernet Header)
  • Smaller MAX_SIZE lower the latency for publishing the virtual channel "pause" status to the remote side
• Use the MAC's FCS for error checking
• In both point-to-point and network mode, using the global ETH pause (IEEE 802.3x) to flow control between the ETH ports on the networkthe Ethernet layer1
  
  • Not using Priority Flow Control (IEEE 802.1Qbb)
• Virtual channel flow control done at the L2 layer

...