Dear all,

Just linked up two sites with 5.5km distances between two of them.
Site A using RB333 with R52H
Site B using RB112 with NMP8602

Throughput test:
RB333-R52H only able to get 13.8Mbps
RB112-NMP8602 able to get 15.1Mbps

Consider the RB333 with R52H is higher spec than RB112-NMP8602, it supposed to be higher output than RB112-NMP8602.
But here are the capture i made: and The winner is

Someone, please analyze this. RB333 with R52H should performed much better.
Its the RouterBoard or the miniPCI of R52H cause it ?

Thanks

The problem is within the RB112. Look at it's CPU occupation: it's always at 100%, so RB333 can't push more than 13Mbps on the other side (RB112), becouse the "other side" is at it's limit. Try RB333+R52H with another RB333+R52H with the same distance/connector/antennas, and you will see higher throughput.

The strange thing is the CPU utilization on RB333: 56% is quite high@13Mbps, i expect max 20-25% of utilization IMHO.

Thanks, I'll try to test RB333-R52 for both side this afternoon.
But i'm still curious, because the test i did @ RB112 side only.
Tested by doing sending and receive.

CPU utilize @ RB112 always reach 100% all the time i did the test, i think it will be to hard to send packets to RB333 as much as 15.1Mbps right ?

Mainly the sending part of bandwidth-test is the thing that consumes a lot of CPU power. So the poor little RB112 is more than at it's limit with creating (!) that much traffic. The better test always is to run bandwidth test from high power machines (2 GHz CPU upwards) behind the routers creating the wireless link...

RB333-R52H at both side.

The result is poor.
Only able to reach max 13Mbps.
The test scenario is the same. Tested on side where RB112 replaced with RB333.

Its time going back to NMP again

Which is the amount of CCQ and bitrate during your test?
Can you made the test without using the bandwidth server onboard but two pc directly connected to the RB333?