Formerly I tested RB333 on my desk to see how it works (I used 2 RBs to create a link and measured the speed of data using Bandwidth Test). Now I tested 2 RB600 and to my surprise the more powerfull RB600 showed worse results in my tests than the RB333. Is there anybody who tried to compare different RB versions?
My test scenario:
2 RBs each with 1 WLAN,
WLANs interconnected via a cable and a attenuator (and Tx power lowered to 5dBm) using 5500 Mhz channel
NSTREME enabled with ‘best fit’
EoIP tunnel between teh RBs made through WLANs (tunnel IPs on WLAN1 interface)
BW test client started on wifi link client side
And my results:
RB333c 3.0rc7
TCP receive: 34mbps (100/90)
TCP send: 34mbps (90/100)
TCP send/receive: 18mbps (100/10)
the numbers in parenthesis means CPU usage (AP/client)
RB600 3.0
TCP receive: 25mbps (100/100)
TCP send: 25mbps (100/100)
TCP send/receive: 15mbps (100/100)
the numbers in parenthesis means CPU usage (AP/client)
The signal quality was 95%-100% in both directions…
So it looks for me like the bigger RB is slower than the smaller one. It sends data at lower rate and it has more loaded CPU during tests. The UDP tests were better for RB600 but it is not what I am interested for too much.
Is the RB600 really slower than RB333 or the different ROS version can make such big difference?
Note: RB333 shows that they have e300c2 CPU running at 333Mhz and RB600 shows that it has e300c1 at 399Mhz. http://www.routerboard.com shows in product table different frequancies and different CPUs…
you should upgrade your RB333 to v3.0
Also you should youse the same wireless cards, pigtails,antennas. Also the location of the routers should be the same, and of course the configuration should be the same.And only then you con compare the test results.
What speeds you get with UDP protocol?
First of all I have to say that I compared the test results again and I found that I compared results from different tests (i.e. WiFi speed without EoIP from RB333 and Wifi with EoIP from RB600 tests). I am very sorry for the confusion - I had an impression that the problem is somewhere outside the RBs.
The RB333c boards in our tests were able to send data over wifi with EoIP tunnel at 24mbps (14mbps if data was sent in both direction simultaneously). So the RB600 with 25.6mbps average speed (15.5mbps in both directions) is slightly better.
Note: Of course, the CPU load will be lower if the bandwidth test will not be run directly on MT (for example using FTP transfer from PCs attached to the MT, etc). I am using BW on MT just to see how much CPU is used - it gives me a picture of what the power of the board is. I wish I have a small RB board which will be able to send data in this test at full rate and the CPU will show 100% of usage
you should upgrade your RB333 to v3.0
I have no free pair of RB333 boards now - in other case I did another test with fresh firmware, of course…
Also you should youse the same wireless cards, pigtails,antennas. Also the location of the routers should be the same, and of course the configuration should be the same.And only then you con compare the test results.
I know that it is not possible to have exactly the same conditions during the tests. But the Wifi cards were the same vendor/type (not the same pieces of course). The cards were interconnected by a cable using an attenuator and Tx power lowered to 5dBm (on RB600 and on RB333). The cables/pigtails were not the same pieces.
Since the quality was 100% or slightly less (for short time) and the speed was consistent without fluctuation and the cards showed that highest modulation (54mbps) is used I think the tests results are comparable and reproducible.
What speeds you get with UDP protocol?
I don’t remember it exactly but it was something slightly over 60mbps (something which seemed to exceed the max speed of Wifi
Could you tell me what attenuator you used for this test please? I am trying to recreate what you have done but am getting erratic ping times cause i have not used an attenuator.
We bought some attenuators from a man who builds antennas for us (very good ones). They are designed for 2.4Ghz but worked in 5Ghz too (of course the attenuation level is unknown for this band). The attenuators we have are not available on market (I think) more…
I just wanted to connect the 5Ghz radios directly via a cable to have stable conditions and no RF power radiated in air (I am having headaches when 5Ghz radio is transmitting in office) an not to overload receivers..
I think you should get link with very good quality using low gain antennas and lowered Xmit power on radios too. Or perhaps you can use long cable between the radios…
