Benchmarking the new WAR board. Shouldn't this thing be faster?

[VerilanES]

I'm configuring 2 WAR boards for a short, fast link to provide streaming video service. This is a very basic deployment, consisting of a single PC on either side of the link. No fancy routing. No multiple connections. I am currently running brand new equipment with all current firmware and apConfig software.

Server:
Super A/G is off
InterBSS Relay is on
802.11a Turbo mode is selected
Access Point mode

Client
Super A/G is off
InterBSS Relay is on
802.11a Turbo mode is selected
Wireless Client Bridge

On both systems, the WPCI and Ethernet adapters are bridged together. I am not using any static routes or RIP. Both atheros cards are also be allowed to scan for the best channel and both are setup to use diversity.

When I run a throughput test using the apConfig utility, I get an average of about 30mb/s. When I run a throughput test using QCheck between two windows boxes behind either side of the link, I'm getting a steady 27mb/s.

I am in an office environment with each side of the setup equiped with 2 P2P 19dbi panel antennas seperated by about 25 feet and pointed right at each other.

I understand that this is a bit close, but I get the same results down a 45 foot hallway. Also, I've set a static rate, static channel, turned on Super A/G, run just a single antenna, used RIP, and always gotten about the same speeds.

Now, I understand that 27mb/s isn't bad, but these speeds are at the very bottom of my deployment model requirements. My understanding is that I should be seeing the full potential of the atheros CM9 cards using the WAR board. In fact, I purchased these boards to replace the limited WRAP boards just so I could get what the forums have advertised as speeds that are "4X faster then the WRAP boards".

I'm open to any and all suggestions. I've got to send this stuff out the door today, but if I could have some suggestions for when my customer calls me pissed off on Monday because his isn't getting at least a stable 35mb/s, I would like to have some sort of answer.

Colin

[lonnie]

What frequency are you using?

[VerilanES]

I've tried it with channel 160, 152 and auto detect mode. All 3 yeilded the same result.

[lonnie]

Can you try it near the bottom and also try without Turbo? The lower channels are for indoor and would not have nearly as much noise on them. Turbo would require a very clean spectrum to get top levels. In most places that is getting near to impossible to achieve. You would be surprised at the amount of 5.x GHz noise already and it is about to get worse since most people are still stuck on this stupid notion that more power is better.

<RANT ON>
There was a thread on one of the lists recently where the guy has these units that are too much power and are overloading his AP front end. The suggestions were to change the antenna to a low gain wide beam --> in effect raising the noise floor on a much wider path for everyone else. They also suggested that he rotate it 90 degrees. Some guys said to aim it high and send the signals into space. Not one guy said the correct thing --> replace the unit with a lower power unit.

Sub 100 mW is all that anybody needs, if they do everything right. For instance my laptop talks to one of our towers about 1 mile away, from inside my building. We have calculated that the signal is going through 6 walls. The connection is not great, but if I step outside it is rock solid. This is a low power radio using the laptop's builtin antenna. Why on earth would someone figure they need a 200 mW radio and 15 to 19 dB antenna to make a connection like that? But they do every day and in steadily increasing numbers. 200 mW and 19 dB antennas are capable of connections to 40 miles. That means you spread your noise 40 miles and more, for the next guy to deal with, with that combination. That is truly stupid and ignorant if all you are trying to achieve a 4 mile connection.
<RANT OFF>

Using IPERF from the remote system to my PC I achieve 44 mbps UDP through a WAR backbone with 4 radios in the loop --> through a midpoint repeater. They are using standard mode, SuperAG features enabled and signal levels of 25+ for quality. The total distance is only a mile, but we use it for throughput testing and it avoids some nasty trees connecting my house if I go direct. We had that direct link using horizontal polarity and it was fine but only had 15 dB quality at the best.

[Skaught]

I agree with your rant.

Our plan for our 5ghz multipoint is CM-9 16dbi sectors and CM-9 CPEs with 19 dbit and 26 dbi antennas. Only thing is we are looking for about 50km range on a 26dbi cpe antenna.

I am not sure that is going to be attainable. I want at least 30km range as that is more or less what we have now on 2.4 with 15dbi client antennas.