On Thu, Apr 15, 2010 at 10:34 PM, RickG <rgunder...@gmail.com> wrote: > Thats leads me to a question. I note RB's website specs for the 450G says: > Actual tested throughput Ether1 <-> Ether2 = 1Gbps
For large-large-large packets, as these boards are pps limited. > Ether2 <-> Ether3 = 650Mbps 1Gbps throughput on ports 1-2 That's because Ether 2 to 5 are connected to a single gigabit CPU port. It should read 500 Mbps and not 650 Mbps, as 650 Mbps would imply a 1.3Gbps port. > Is the 750 the same? THe RB750 don't use the RB450G and RB750G switch chip. (http://wiki.mikrotik.com/wiki/Manual:Switch_Chip_Features) > Also, what does "optional Switch Chip functionality for wire speed > Gigabit throughput." mean? It means if you configure switching instead of CPU-forwarding on these ports, they will get wire speed throughput. But it will be limited layer-2 and may be tag insertion/removal, some L3/L4 ACLs if they are small. > This is interesting as well: > "Comparing to RB750, the G version adds not only Gigabit capable ports, but > a new 680MHz Atheros 7161 CPU for increased throughput. Up to 580Mbps > throughout with larger packets, and up to 91500pps with small packets!" According to the page above RB750G doesn't have the "all-port-switch" option of RB450G, which suggests it only has one gigabit connection to the CPU. The fact that all RB750G ports have the same MTU (http://wiki.mikrotik.com/wiki/Manual:Maximum_Transmission_Unit_on_RouterBoards), when RB450G Ether1 has a slightly larger MTU than RB450G Ether2-5 suggests that as well. Rubens -------------------------------------------------------------------------------- WISPA Wants You! Join today! http://signup.wispa.org/ -------------------------------------------------------------------------------- WISPA Wireless List: wireless@wispa.org Subscribe/Unsubscribe: http://lists.wispa.org/mailman/listinfo/wireless Archives: http://lists.wispa.org/pipermail/wireless/