Page 11 of 12

Re: ARM devices and NV2 protocol

Posted: Sat Feb 23, 2019 1:08 am
by djvolt
This is not AC mode, AC mode will reach around 240Mbps on 40MHz :P

Re: ARM devices and NV2 protocol

Posted: Sat Feb 23, 2019 4:46 am
by OniLink
Hi, then is ARM already fixed? There are no more packet losses?

Re: ARM devices and NV2 protocol

Posted: Sat Feb 23, 2019 8:39 am
by 2jarek
This is not AC mode, AC mode will reach around 240Mbps on 40MHz :P
yeee single TCP stream 240 Mbit from 40 Mhz :D Mby in lab.

Re: ARM devices and NV2 protocol

Posted: Sat Feb 23, 2019 9:03 am
by ste
This is not AC mode, AC mode will reach around 240Mbps on 40MHz :P
yeee single TCP stream 240 Mbit from 40 Mhz :D Mby in lab.
Still a waste of spectrum these days. Hope .ax shows up soon. >100M Capacity/10MHz is 2019.

Re: ARM devices and NV2 protocol

Posted: Sat Feb 23, 2019 9:34 am
by server8
With ubiquiti AF HD we are very close to have 100 mb/s download speed with 10 MHz for PtP NOW for AP side with cambium we have >300 Mb/s@20 MHz using ultra dense mu-mimo technology

Re: ARM devices and NV2 protocol

Posted: Sat Feb 23, 2019 9:52 am
by human1982
With ubiquiti AF HD we are very close to have 100 mb/s download speed with 10 MHz for PtP NOW for AP side with cambium we have >300 Mb/s@20 MHz using ultra dense mu-mimo technology
So if DISC Lite5 ac DiscG-5acD 240mbit (if it really works) cost is 45$, what is the cost of cambium per mbit?
This is NV2 ARM thread.

Re: ARM devices and NV2 protocol

Posted: Sat Feb 23, 2019 10:04 am
by ste
With ubiquiti AF HD we are very close to have 100 mb/s download speed with 10 MHz for PtP NOW for AP side with cambium we have >300 Mb/s@20 MHz using ultra dense mu-mimo technology
So if DISC Lite5 ac DiscG-5acD 240mbit (if it really works) cost is 45$, what is the cost of cambium per mbit?
This is NV2 ARM thread.
Wisps learned that the expensive resource is spectrum not Hardware. To stay on topic: NV2 Arm has to be improved much further to be more spectrum efficient.

Re: ARM devices and NV2 protocol

Posted: Sat Feb 23, 2019 10:05 am
by 2jarek
With ubiquiti AF HD we are very close to have 100 mb/s download speed with 10 MHz for PtP NOW for AP side with cambium we have >300 Mb/s@20 MHz using ultra dense mu-mimo technology
Omg 2x2 mimo vs 4x4 mimo compare No sense. AF HD 2x2 true but for 4096 QAM nede clean spectrum

Re: ARM devices and NV2 protocol

Posted: Sat Feb 23, 2019 6:05 pm
by sergesa
ìt really work ! PtP 20MHZ TCP
1. 802.11 105/100
2. NV2 90/80
nv2 work what never before on ARM, but we wait continued improvements
This still shows 802.11 is better?
What signal strength on both ends? 1 stream tcp or more? Test between the two devices or from devices behind the two radio's? (like two CCR behind the 'arm' devices)
Rudy,
as my result yes, 802.11 is better, but in ptp.. I'm not sure why I need to use TDM in ptp configuration. I wait this mainly for PtMP, and now will try to update and make test (ap under load all time)
I think now is not matter use external bandwidth test, because last MT firmware use all cores for test
20 streams by default. Bellow is my test PtP

Re: ARM devices and NV2 protocol

Posted: Mon Feb 25, 2019 10:11 am
by mfr476
at the end works, a little slow but work¡¡¡¡¡¡

Re: ARM devices and NV2 protocol

Posted: Mon Feb 25, 2019 12:27 pm
by joserudi
Now that works perfectly nv2. Will it release mikrotik mantbox with arm that admits a greater number of users and throughput?

Re: ARM devices and NV2 protocol

Posted: Mon Feb 25, 2019 12:34 pm
by mfr476
I think first mikrotik release omnitik ac arm but i don't know

Re: ARM devices and NV2 protocol

Posted: Mon Feb 25, 2019 1:29 pm
by server8
Now it is much better but it sometimes miss the NV2 slot and the modulation drop.

When I stop the speedtest from the arm device the tx and rx modulations goes higher, the speedtest using UDP is perfect using TCP unleashes more the issue the cpu values are:

system resource cpu print follow
# CPU LOAD IRQ DISK
0 cpu0 5% 0% 0%
1 cpu1 33% 33% 0%
2 cpu2 9% 0% 0%
3 cpu3 3% 0% 0%

Here a print screen left mipsbe AC and left arm AC both clients are on the same roof the signal is better on the arm device

Now that works perfectly nv2. Will it release mikrotik mantbox with arm that admits a greater number of users and throughput?

Re: ARM devices and NV2 protocol

Posted: Mon Feb 25, 2019 1:42 pm
by mfr476
Lite to lite

Re: ARM devices and NV2 protocol

Posted: Mon Feb 25, 2019 7:49 pm
by server8
with v6.44rc4 seems to work much much better MT is very close to solve the iusse

Re: ARM devices and NV2 protocol

Posted: Mon Feb 25, 2019 10:12 pm
by honzam
with v6.44rc4 seems to work much much better MT is very close to solve the iusse
Your picture is with RC1 (this is without ARM fix)

Re: ARM devices and NV2 protocol

Posted: Tue Feb 26, 2019 8:55 am
by normis
server8 you are not testing any ARM fixes. The fixes are in rc4 only.

Re: ARM devices and NV2 protocol

Posted: Tue Feb 26, 2019 9:03 am
by server8
I have posted that with RC4 works much much better the performance now are much much closer to old AC devices :-)

Re: ARM devices and NV2 protocol

Posted: Tue Feb 26, 2019 10:01 am
by 2jarek
RC4 for ARM works much much better for NV2 & pure 802.11 too. Here MIPSBE AC Basestation + ARM CPE good signal 50 dB but non line of sight for pure 802.11N mode. Before RC4 latency spikes for idle mode / no traffic only.

Re: ARM devices and NV2 protocol

Posted: Tue Feb 26, 2019 10:39 am
by honzam
The fixes are in rc4 only.
Normis, is there diference between rc4 and final 6.44? In wireless driver? Thanks

Re: ARM devices and NV2 protocol

Posted: Tue Feb 26, 2019 10:49 am
by emils
6.44rc4 and 6.44 versions are identical.

Re: ARM devices and NV2 protocol

Posted: Tue Feb 26, 2019 10:23 pm
by wispman74
just a question; now for ptp link wich is better between nv2 and 802.11n?

Re: ARM devices and NV2 protocol

Posted: Tue Feb 26, 2019 11:03 pm
by networkfudge
just a question; now for ptp link wich is better between nv2 and 802.11n?
nv2 was designed to optimise performance of point to multipoint networks. Usually vanilla 802.11 is faster in ptp however in cases where there is interference nv2 can perform better than 802.11.
When setting up ptp links I always cycle through 802.11/nstreme/nv2 and pick the best. Quite often this is 802.11 or nstreme and not nv2, but when setting up wireless links every environment is different so test test test and don't rely on what worked for someone inside a forum somewhere. Ptmp I always use nv2 though lately I have been hearing that 802.11 is performing better than it used to.

Re: ARM devices and NV2 protocol

Posted: Tue Feb 26, 2019 11:58 pm
by WirelessRudy
just a question; now for ptp link wich is better between nv2 and 802.11n?
nv2 was designed to optimise performance of point to multipoint networks. Usually vanilla 802.11 is faster in ptp however in cases where there is interference nv2 can perform better than 802.11.
When setting up ptp links I always cycle through 802.11/nstreme/nv2 and pick the best. Quite often this is 802.11 or nstreme and not nv2, but when setting up wireless links every environment is different so test test test and don't rely on what worked for someone inside a forum somewhere. Ptmp I always use nv2 though lately I have been hearing that 802.11 is performing better than it used to.
Don't forget to mention that even P2P links once set need every so many months a new check. Spectral environment change. Other users (operators) start using the same or near frequency for instance... A link once worked fine can be a disaster half a year later...

I am one of the user that for the last half year show many tests where 802.11 (rts/cts) outperforms NV2 almost every time again. And special in 'ac' the technology advanced where NV2 in my opinion did not....

Re: ARM devices and NV2 protocol

Posted: Wed Feb 27, 2019 2:25 am
by adilsemedo
6.44rc4 and 6.44 versions are identical.
For the accurate test, does the AP (netmetal) should also has the 6.44 version or doest it matter?

Re: ARM devices and NV2 protocol

Posted: Wed Feb 27, 2019 5:25 pm
by Alessio Garavano
6.44rc4 and 6.44 versions are identical.
For the accurate test, does the AP (netmetal) should also has the 6.44 version or doest it matter?
Brother, the AP is the main device to be upgraded!

Re: ARM devices and NV2 protocol

Posted: Thu Feb 28, 2019 9:27 am
by adilsemedo
Hi all,
Someone has tested more scenarios with ARM, 5ghz AC, ROS 6.44 with NV2 or 802.11?
we´ve been facing this issues on our network and we need to solve this...

~150 RB922 AP and ~2000 arm CPE

Re: ARM devices and NV2 protocol

Posted: Thu Feb 28, 2019 9:28 am
by normis
please upgrade and see. most people have reported good results.

Re: ARM devices and NV2 protocol

Posted: Thu Feb 28, 2019 9:34 am
by networkfudge
Hi all,
Someone has tested more scenarios with ARM, 5ghz AC, ROS 6.44 with NV2 or 802.11?
we´ve been facing this issues on our network and we need to solve this...

~150 RB922 AP and ~2000 arm CPE
Upgrade yesterday!!!!!!!!!!!

Re: ARM devices and NV2 protocol

Posted: Thu Feb 28, 2019 10:34 am
by server8
We are making some tests and seems to work much much better but our scenario is 1 arm client and 30 mipsbe clients on the AP

I don't know if more arm clients on the AP is a problem

Re: ARM devices and NV2 protocol

Posted: Thu Feb 28, 2019 12:10 pm
by adilsemedo
We are making some tests and seems to work much much better but our scenario is 1 arm client and 30 mipsbe clients on the AP

I don't know if more arm clients on the AP is a problem
@Server8,

wich MIPSBE device do you use as clients? I saw at this thread that mispse is betther than arm for AC...
The only have DynaDish5, the newer SXT AC lite is all arm...

Re: ARM devices and NV2 protocol

Posted: Thu Feb 28, 2019 2:10 pm
by server8
Old SXT lite AC no more on the market actually we are installing mipsbe N after arm ac NV2 issue, yesterday we start back to install some arm ac devices to check it in real life but it's soon to understand if they works

Re: ARM devices and NV2 protocol

Posted: Thu Feb 28, 2019 6:46 pm
by InoX
nstreme will be ditched? Or is it allready?

Re: ARM devices and NV2 protocol

Posted: Thu Feb 28, 2019 6:58 pm
by WirelessRudy
please upgrade and see. most people have reported good results.
Ok, yesterday upgraded a NetMetal5 from 6.34.12 to 6.44 and all its 9 clients (a mixture of Sextants, LHG's and SXT lites) the same. All mipsbe and all working on 'n' modulation.
Although the NV2 improved a lot, 802.11n with rts/cts is still some 30% faster. The AP and its clients run in a very congested spectrum. Can find no 'free' channel in 40Mhz, 20Mhz sees adjacent AP's hitting with -70 etc. so although we run 40Mhz bandwidth, it sees some overlap with some -75 AP at distance. 802.11 has no issue to maintain almost consistend 50Mbps download towards clients, NV2 stays around 35Mbps. But that was no more then 25 so yes it improved... (or is spectrum different today?)

Yesterday als ran one Omintik5-ac with only 4 clients, all arm units (except this Omnitik off course) and here we saw that NV2 now performs almost the same as the 802.11 rtx/cts protocol.

During the week that comes I will test more P2MP networks since overal I do not see new issues so I can only expect to gain by upgrading.....

Re: ARM devices and NV2 protocol

Posted: Thu Feb 28, 2019 7:57 pm
by adilsemedo
Now i´ve an AP with 26 ARM CPE. I wounder if i can do a stress test to all units at the same time, like high BW.

Re: ARM devices and NV2 protocol

Posted: Thu Feb 28, 2019 7:59 pm
by WirelessRudy
This is not AC mode, AC mode will reach around 240Mbps on 40MHz :P
yeee single TCP stream 240 Mbit from 40 Mhz :D Mby in lab.
Impossible:
802.11ac MCS9 = 200Mbps is highest possible connection rate for 40Mhz channel, one stream. You are on dual stream. You can't beat the standard...

Re: ARM devices and NV2 protocol

Posted: Thu Feb 28, 2019 8:02 pm
by WirelessRudy
Now i´ve an AP with 26 ARM CPE. I wounder if i can do a stress test to all units at the same time, like high BW.
You can, but you'll have to login to every unit and run it. You need a lot of open windows in your management PC..... And you'll find the maximum throughput is relatively low since the AP needs to process all that traffic for each unit which creates a lot of overhead that is bringing the usable throughput down.
Better is to stress test with 3-4 clients max, which is probably also much more reflecting reality....

Re: ARM devices and NV2 protocol

Posted: Fri Mar 01, 2019 12:30 pm
by marcin21
Are there any improvements in nv2 regarding older architecture, mipsbe+AC (omnitik ac) ?
I'm still using 802.11 mode in such scenario and for some longer period of time I stopped buying those omnitiks.

Re: ARM devices and NV2 protocol

Posted: Sat Mar 02, 2019 12:39 am
by peson
Now i´ve an AP with 26 ARM CPE. I wounder if i can do a stress test to all units at the same time, like high BW.
You can, but you'll have to login to every unit and run it. You need a lot of open windows in your management PC..... And you'll find the maximum throughput is relatively low since the AP needs to process all that traffic for each unit which creates a lot of overhead that is bringing the usable throughput down.
Better is to stress test with 3-4 clients max, which is probably also much more reflecting reality....
Well you only need to login into one router, a core like a CCR is prefered.
Open up some terminal windows and run:
/tool bandwidth-test 1.2.3.4 user=admin password=password protocol=udp local-tx-speed=5M remote-tx-speed=2M direction=both
Do this for as many CPEs you would like to test, select different speeds for some and you will have a good test of your network.

Re: ARM devices and NV2 protocol

Posted: Sat Mar 02, 2019 1:59 am
by WirelessRudy
Now i´ve an AP with 26 ARM CPE. I wounder if i can do a stress test to all units at the same time, like high BW.
You can, but you'll have to login to every unit and run it. You need a lot of open windows in your management PC..... And you'll find the maximum throughput is relatively low since the AP needs to process all that traffic for each unit which creates a lot of overhead that is bringing the usable throughput down.
Better is to stress test with 3-4 clients max, which is probably also much more reflecting reality....
Well you only need to login into one router, a core like a CCR is prefered.
Open up some terminal windows and run:
/tool bandwidth-test 1.2.3.4 user=admin password=password protocol=udp local-tx-speed=5M remote-tx-speed=2M direction=both
Do this for as many CPEs you would like to test, select different speeds for some and you will have a good test of your network.
Ok, interesting. Never new that.... going to try that one day... now its weekend... But thanks anyway!

Re: ARM devices and NV2 protocol

Posted: Sat Mar 02, 2019 2:02 pm
by pr0fil
SXTsq 5 AC(arm)
ptp Link 80Mhz
using nv2
Throughput - 450mb/s
no packet loss
distance ~600m
looks very good, nv2 is best case scenario atm.

Re: ARM devices and NV2 protocol

Posted: Sun Mar 03, 2019 12:08 pm
by networkfudge
This is not AC mode, AC mode will reach around 240Mbps on 40MHz :P
yeee single TCP stream 240 Mbit from 40 Mhz :D Mby in lab.
Impossible:
802.11ac MCS9 = 200Mbps is highest possible connection rate for 40Mhz channel, one stream. You are on dual stream. You can't beat the standard...
The guy never said he was SISO, that was your own assumption. He said SINGLE TCP STREAM, with respect to the bandwidth test.
His calculations (240mbps @ 40mhz) are still incorrect though at least when it comes to PHY rates.

Perhaps 240mbps throughput is his real world experience in the field with 2-stream AC @ 40mhz, but he hasn't really explained what he means so we can only guess!

Re: ARM devices and NV2 protocol

Posted: Sun Mar 03, 2019 11:35 pm
by adilsemedo
Now i´ve an AP with 26 ARM CPE. I wounder if i can do a stress test to all units at the same time, like high BW.
You can, but you'll have to login to every unit and run it. You need a lot of open windows in your management PC..... And you'll find the maximum throughput is relatively low since the AP needs to process all that traffic for each unit which creates a lot of overhead that is bringing the usable throughput down.
Better is to stress test with 3-4 clients max, which is probably also much more reflecting reality....
Well you only need to login into one router, a core like a CCR is prefered.
Open up some terminal windows and run:
/tool bandwidth-test 1.2.3.4 user=admin password=password protocol=udp local-tx-speed=5M remote-tx-speed=2M direction=both
Do this for as many CPEs you would like to test, select different speeds for some and you will have a good test of your network.
Ok, interesting. Never new that.... going to try that one day... now its weekend... But thanks anyway!
Thanks so much @peson. Very good material,
i´ll test it tomorow.

We have a virtual machine, running Dude Server at the core network, where we can do this test scenario:
But if we have 4 CPEs with different management IP (ex: 10.10.10.1, 10.10.10.2, 10.10.10.3 and 10.10.10.4), we have to running 4 sessions of:
/tool bandwidth-test 10.10.10.x user=admin password=password protocol=udp local-tx-speed=5M remote-tx-speed=2M direction=both

Re: ARM devices and NV2 protocol

Posted: Mon Mar 04, 2019 1:36 am
by OniLink
SXTsq 5 AC(arm)
ptp Link 80Mhz
using nv2
Throughput - 450mb/s
no packet loss
distance ~600m
looks very good, nv2 is best case scenario atm.
hi, can you share speedtest screenshots between radios and ping? please friend

Re: ARM devices and NV2 protocol

Posted: Mon Mar 04, 2019 10:39 am
by pr0fil
UDP + torrent, i can share with TCP if you need.

Re: ARM devices and NV2 protocol

Posted: Tue Mar 05, 2019 11:03 am
by OniLink
UDP + torrent, i can share with TCP if you need.

thank you very much for sharing, greetings

Re: ARM devices and NV2 protocol

Posted: Sun Mar 10, 2019 3:28 pm
by alejosalmon
Hello is there any improvement in nstreme in routeros 6.44?

Re: ARM devices and NV2 protocol

Posted: Sun Mar 10, 2019 4:32 pm
by honzam
Hello is there any improvement in nstreme in routeros 6.44?
No, only NV2

Re: ARM devices and NV2 protocol

Posted: Sun Mar 10, 2019 4:48 pm
by DavidAlves
Hi! I would like know about nv2 on mipsbe. Is the performance was improved too or better stay on 6.42? On my network we never get over of about of ~50mb on nv2@20mhz on rush hour and about 25 client with plans between 5mb ~10mb.

Thanks!
David

Enviado de meu Mi A2 Lite usando o Tapatalk


Re: ARM devices and NV2 protocol

Posted: Tue Mar 12, 2019 12:47 pm
by 2jarek
@20 MHz 802.11N (not AC) NV2 now 110 Mbit/s for TCP speedtest good job mikrotik :D . Now time for fix 40 MHz channel in this moment max speed only 150 Mbit for TCP :(