bullet M5 performance help!

[rb384997]

Mike,

40Mhz Channel width, shifting enabled
airmax on, aggregation on
transmission rate best, auto
I have attached main page stats of ap and cpe...
mikrotik to mikrotik bandwidth test:

(Same time)
12Mbps send
12Mbps Receive

if i turn air max off i can do 30Mbps one way
if i test send and receive at the same time I get
about the same test results with airmax on,12 Mbps send and 12Mbps receive.....

What is the deal? Can you help tweek my settings?


Oh and i updated to your new beta firmware with identical test results...

Thanks,

RC

[UBNT-Mike.Ford]

Mike,

40Mhz Channel width, shifting enabled
airmax on, aggregation on
transmission rate best, auto
I have attached main page stats of ap and cpe...
mikrotik to mikrotik bandwidth test:

(Same time)
12Mbps send
12Mbps Receive

if i turn air max off i can do 30Mbps one way
if i test send and receive at the same time I get
about the same test results with airmax on,12 Mbps send and 12Mbps receive.....

What is the deal? Can you help tweek my settings?


Oh and i updated to your new beta firmware with identical test results...

Thanks,

RC


hello,

Please turn off channel shifting and test again. Also how many streams do you have the MikroTik tester running?

Thanks,

[lncommunications]

Use jperf! Mikrotik btest does not work properly, you can follow my result postings in the 2.4 M Series Product forum.

Or

http://www.ubnt.com/forum/showthread.php?t=12740

[rconaway]

30Mbps with a 20MHz channel and single transfer is about right but we were testing early firmware. We will retest right now.

[drewlsvern]

30Mbps with a 20MHz channel and single transfer is about right but we were testing early firmware. We will retest right now.

I thought these were capable of 100 mbps?

[rconaway]

It's how you do the testing. This product is different that many others. For some reason, if you do a computer to computer raw data transfer such as an ftp transfer or a single stream transfer, 30Mbps to 50Mbps is about what you get, Bullet to Bullet. However, have multiple computers do one way ftp transfers and the rate goes up signficantly. My personal belief is that there is still some way to go in firmware that will improve this. Based on our testing and monitoring CPU overhead, certain types of traffic were being handled at different CPU rates. However, for a single stream, CPU overhead ranged from 40-60%. That shows me that the processor wasn't maximized on throughput yet.

[Dave-D]

Rory, what's the quickest way to
monitor CPU loading? Dave

[rconaway]

Good question. I can't find my notes on this. If someone has this info, please jump in while I look.

[drewlsvern]

It's how you do the testing. This product is different that many others. For some reason, if you do a computer to computer raw data transfer such as an ftp transfer or a single stream transfer, 30Mbps to 50Mbps is about what you get, Bullet to Bullet. However, have multiple computers do one way ftp transfers and the rate goes up signficantly. My personal belief is that there is still some way to go in firmware that will improve this. Based on our testing and monitoring CPU overhead, certain types of traffic were being handled at different CPU rates. However, for a single stream, CPU overhead ranged from 40-60%. That shows me that the processor wasn't maximized on throughput yet.

not trying to hijack this thread, but what about the bullet m2 and nanostation m5 do they work the same way as well? What throughput are they capable of?

Thanks
Joel

[UBNT-Mike.Ford]

Also,
Please see my test plan for testing 802.11n at the front of this forum. It shows how to fully saturate an 80.11 link using a bandwidth tester.

Also test 2 windows machines using Windows file transfer and you generaly wont see more then 50Mbps transfer rate anyways. Windows isnt the most efficient networking tool out there.

Thanks,

Mike