I encounted a frequency of 161.1328125Mhz recently, and I digged a bit about its origin (because the figure seems odd at first glance). It seems that for 10GbE standards, it uses 64/66b encoding, so to achieve 10Gbps, the signal baudrate has to be 10*66/64=10.3125(Gbaud). And 10.3125G/64=161.1328125M, so this gives a reason for 161.1328125M that it's a base freqeuncy for 10GbE signals. And there are 161.132M crystal oscillators used in production as I see some on my PCIe cards.
For 100GbE, it seems that the same base frequency pops up again: it's said that 100GbE uses 256/257b encoding with 514/528 RS-FEC, so the signal baudrate has to be 100*257/256*528/514=103.125(Gbaud), which is 10 times the baudrate of 10GbE. So the same base frequency 161.1328125MHz can be used.
Actually 64/66 is exactally equal to 256/257*514/528. What a coincidence! I am guessing that, when people were choosing the encoding/FEC scheme for 100GbE, they deliberately chosen these numbers (256/257 and 514/528) such that baudrate of 100GbE is 10 times that of 10GbE (thus the 161.1328125M crystal oscillators can also be used).
Does my guess make sense?
