Hi
I managed to get amazing transmission results results with 3x3 hardware.
Here is sample:
Nice!
Nice!!
What radio use? board? What kind of test use?
I do test with iperf half duplex TCP and I can reach 300mbps
It was tested with MT Btest on Atom motherboard.
Here is a full review. It will be translated to english, but now it’s only available in polish.
http://www.cyberbajt.pl/raport/475/
What board are you using?
I was leaning towards an Atom board just for mini PCIe, still come out cheaper than an RB800.
What kind of antenna do you used?
I’m using this antennas:
http://www.cyberbajt.com/en/shop_item/3324/pan_519mimo_3x3.html
More info is on their polish site:
http://www.cyberbajt.pl/produkt/4332/
I’m using Intel DN2800MT motherboard:
Here is a comparison of motherboard perfomance:
Ehm, your result table is completely off - i just run simple test Trough my RB493G - that have some configuration on it and with single TCp connection to my linux server i got more than 300Mbps throughput. I’'m sure if i will clear up configuration and will make more TCp connections i will get even more.
Only way i can see you can get these results is if you run test to or from the router or run loopback to itself.
Loopback results are meaningless, they only show that the ATOM has more CPU Cache. See what Macgaiver said above. Correct TCP test procedure will give good results on RouterBOARD devices.
OK - I will check it.
Yes - it’s loopback test. On the router I make BTest to loopback (127.0.0.1)
You did test to linux platform so you had to use Ethernet connection. RB493G has a switch port.
If you do this you are not testing the processor of the platform it self.
Do fdx TCP test to your linux server and post the results.
You did test to linux platform so you had to use Ethernet connection. RB493G has a switch port.
If you do this you are not testing the processor of the platform it self.
Do fdx TCP test to your linux server and post the results.
It was routing test, so all traffic went trough CPU. port–switch1–CPU–switch2–port to be exact.
it was FTP download.
I think it may interest some forum users:
https://www.cyberbajt.com/en/news/6/0/mimo_3x3.html
Interesting tests. Here is our test of 3x3 MIMO on RB800:
Taken from here: http://mum.mikrotik.com/presentations/PL12/workshop-wireless-2012-PL.pdf
Like macgaiver pointed out above, TCP loopback test is very misleading and doesn’t show actual performance. You would need to test through the device with actual traffic, or iperf.
Further on in the article they show that doing this actually gives good results, as opposed to loopback test.
Sory Normis, but Btest results are for me meaningless. We made some p2p 2+2Mimo links showing Btest results over 200Mbits but the normal trafic was around 50-70Mbit, using 1+1chain Btest shows 100+Mbits and real performance 90+Mbits …
So right now we are preparing brand new p2p link, using the brand new Jirous JR S24-MIMO antenas
, using brand new 433GL and brand new R52Hn cards, so lets see how it ends.
Ill report the results.
This is the best hardware what i can use from Mikrotik on this place, so i hope for some reasonable performance.
Karlos, what do you mean by “normal traffic”? many programs don’t actually utilize the maximum possible speed. you could test with multiple FTP connections
Yes, to test normal traffic we use just normal PC using normal multiple http or ftp transfers.
We found some places in our network that work very good using nv2 but the same setting wont work on another location. We also have one very strange location unable to use 802.11n at all, but 802.11a works great. This is not always about noisy area, nearly no traffic in the air, but N does not work at all. We changed antenas, boxes, cards, boards, pigtail… nothing. But we have some another places and found that 802.11N using normal modulation works pretty slow, but nv2 improoves transfer rate 3times. It really deppends on location more than another factors.
But like i promissed, ill give results on this brand new p2p link. (Coplete signal levels, settings, speeds, loads, distance etc…) Please wait for it.
Card rates Fixed do 15dBm
Distance 1800m
Signal Strenght -50/-50dBm
CCQ 97/96%
Rate 300/300Mbps
Frequency 5500Mhz
Band 5GHz only N
40MHz width
protocol nv2
Datarates advanced
0+1chain
HT AMPDU 0+4
HT guard ANY
lowering card power will cause lowered transfer rate
more card power will not help but does not hurt the transfer rate or CCQ, its just useless
I also found that the “wanted” power is not the output power of the card. So i started with maximum power (default) and than lowered the output using signal RX strenght on other side of the link. We need to stay under 30dBm (or pretty near to that value)
BTest UDP MK 2 MK direct just over the wirelles 125/105Mbit simultanously and pretty stable. (230Mbit agregate) CPUs still not on 100%load
Normal test using Notebook
One stream http data average 37Mbit max 45Mbit
Five streams http data average 65Mbit max 76Mbit
So is it ok ? Or is something wrong ?
Thank you