Welcome to the wonderful world of 802.11 and marketing.

Basically the 65 mbit is the raw low-level speed, before overhead has been taken into account, and 802.11 has significant overhead.

This is why some alternative technologies seem slower, because those technologies quote the actual speed you will get, rather than the low-level protocol speed.

Welcome to the wonderful world of 802.11 and marketing.

Basically the 65 mbit is the raw low-level speed, before overhead has been taken into account, and 802.11 has significant overhead.

This is why some alternative technologies seem slower, because those technologies quote the actual speed you will get, rather than the low-level protocol speed.

I thought it was more because speed tests expect full duplex operation, and wireless only works in half duplex? Forgive me if I'm wrong.

I suppose there would be some bandwidth "lost" due to layer-3 ack packets having to come back the other direction, but its a pretty small amount, certainly much less than 50%. 802.11 does not reserve bandwidth, so if you have no traffic in one direction, you should be able to get full speed in the other, minus its overhead. Part of that overhead though is that 802.11 has its own layer-2 ack packets.

Many speed tests test one direction, and then the other. The RouterOS speed test can test one direction, the other direction, or both directions simulaneously.

In the case of an 802.11 link, the both directions test should be half the speed of either of the single direction tests. (And if one direction is different than the other direction, then you probably have an interference or signal or power level problem in one direction that is different than the other).