there are a couple things i noticed on the routerboard.com website that don't seem right.

for one, the CCR1009-8G-1S-1S+PC shows on the main product listing page that it is 1.2Ghz, but on the CCR1009-8G-1S-1S+PC's individual page, it says 1Ghz. so which is it?
also, looking at the performance results on the CCR1016-12G, the 1518 byte testing looks to have been just copy/pasted all the way down the line...there's not even a 0.1pps difference between all 5 tests at 1518 byte? that just doesn't seem likely unless there's some single bottleneck that affects the testing in all 5 cases?

1. you can set freq from 400mhz to 1200mhz
2. maybe thats what is max at mikrotik? - it does not bottnenek packet per sec - but speed of interface. look at other CCR's - and ou see almoust the same

yeah, i know you can change the frequency; but the speed that is generally advertised is the nominal "default" speed, how it comes set from the factory. in this case it lists 2 "nominal" speeds in 2 different places. if it comes from the factory at 1Ghz, it should be listed in both places as 1Ghz, and 1.2Ghz in both places if it comes stock at 1.2Ghz.

as far as the packets per second, that's exactly what i mean....the results listed seem to indicate that there is a bottleneck somewhere that is causing the performance to be the same across the board in all the tests, but it doesn't really indicate where that bottleneck is. even still, in actual, real-world testing, it seems highly unlikely that the tests would turn out exactly the same across all 5 tests...it looks as though the numbers were either averaged over multiple tests, and those results happened to be the same, or the results were so close that they just decided to use the same numbers all the way down the list because it was easier. i don't know that this is actually the case, but those are the first thoughts that go across my mind....and that's kinda the point i'm trying to make...we shouldn't be left to guess, they should state somewhere the reason the results are all the same to avoid such questions.

I think the Mbps limit is that 1 packet more of that size would imply more than 1Gbps per interface.

It would be good to also show full duplex numbers in this case.