So I have been working with CapsMan and getting a better feel for it. Not having the scanning tools is a bit of an issue... but lets get down to the real question.

What sort of WiFi performance are you actually seeing?

I just got a 1000/20M connection at my office.

So first we establish a base line. RB3011 Hardwired.


Then we have the hAP AC running as a WAP.


Then we have the hAP AC running as a CAP


So lets Compare it to a UniFi AC Lite


What about the UniFi AC LR


Lastly Ruckus R510


All tests on the same cable using a Ubnt toughswitch as the POE.
Each AP was plugged in one at a time in the same spot.
Each set to Channel 149 80mhz
Same Laptop used in the same spot. (Wireless Card is AC-MU)
The Screen Shots were the average over 5 or 6 tests each.

Additional.

There is this benchmark: http://blog.quindorian.org/2016/02/revi ... reat.html/

What CPU usage are you seeing on the hAP AC in the tests? Are you using local forwarding with CAPsMAN?

Default on the local forwarding. (Off I assume)

CPU during download tests 70-80%

Default on the local forwarding. (Off I assume)

CPU during download tests 70-80%

Using local forwarding with CAPsMAN puts much less load on the device, I would recommend using local forwarding, your speeds should increase.

I would also check the profiler to see what is using it, it sounds like you might be nearing the CPU maximum. You might also fiddle with the hardware queue setting for the wlan interfaces to see if it improves throughput, in case the default SFQ applied to wlan interfaces is putting additional CPU load on the router.

Note, in the review, with iperf, he got 500Mbps transfers when the hAP AC was just being used as an access point, so I would expect close to that. Have you tried testing with iperf?

As a WAP... I did get 340.

But its troubling to see it (Mikrotik) beat buy UniFi--k gear. I have written them off for a list of reason.

Guess I need to break them back out again and run more tests.

I also seem to be stuck with ~ 220mbps, tested with 3x3 802.11ac 5 GHz USB 3 adapter at ~ -60dBm and iperf3. Only one other AP nearby on a far away channel. Not really sure how to get it any faster, no matter what I tweak nothing seems to change. It does seems on the low end compared to other vendors. Ubiquiti always seem to have the best wireless performance, wonder if it has something to do with their earlier years of developing miniPCI radio cards before branching into integrated solutions.

As a WAP... I did get 340.

But its troubling to see it (Mikrotik) beat buy UniFi--k gear. I have written them off for a list of reason.

Guess I need to break them back out again and run more tests.

Exactly around 350Mbit is what i get from hAP AC(connection on laptop shows 780mbit, dual-chain), running it as simple AP nothing more, no special settings, in addition to that when i try to copy around 30gb files from network it breaks around 50% and 5ghz WLAN dies for around 2 mins as i described here http://forum.mikrotik.com/viewtopic.php ... 27#p571527

So theres for sure something wonky around this APS and so far i haven't got any useful information back on the issue...

Yeah I cancelled my order because of this and other tests. It seems there's pretty poor performance due to the CPU bottleneck. Hopefully they'll come out with a new version based on their new HAP since it has a faster cpu.

wAP AC Running as Cap

My test results using a Mac Book Pro (AC 3x3):
http://forum.mikrotik.com/viewtopic.php ... 88#p556472

When using CAPsMAN, everything is processed using the CPU, since CAPsMAN does not support FastPath.

Without using it as a cAP, you can expect around 400Mbit on this device in 5GHz.

@normis, thanks for your reply.
Does MikroTik offer any faster performing 5GHz devices, yet?

When using CAPsMAN, everything is processed using the CPU, since CAPsMAN does not support FastPath.

Without using it as a cAP, you can expect around 400Mbit on this device in 5GHz.

So the RB3011 to a wAP AC throughput at 8 feet looks right?
I have a ccr1009-8g-1s-1s+pc I could flip to if you think that would help???

So the RB3011 to a wAP AC throughput at 8 feet looks right?
I have a ccr1009-8g-1s-1s+pc I could flip to if you think that would help???

Packets from wAPs wireless to ether are not going fasttrack yet because of CAPsMAN. Limiting factor is wAP CPU.

The reduced throughput looks like a Mikrotik bug, even if it might not be.

That's because the TP-Link Archer c7 that's based on the same chipset gets 500 Mbps with 2x2 with some versions of the firmware and driver, and gets the same 340 with other versions, which seems to indicate that the hardware is capable of more. (There's a long thread on openwrt forums measuring performance of that chipset using different ath10k firmware/driver revisions).

I haven't seen a good explanation from Mikrotik as to what in particular is limiting the performance. I also don't know if Mikrotik has access to the firmware running on the card, or are they contracting someone who does, or neither.

TP-link is not running in a CAPsMAN managed network. You can't compare managed network and simple AP situation. If you use hAP ac as a simple AP without any CAPsMAN, it will work at 400Mbit easily. This was mentioned above a few times already.

So, does MikroTik offer any device that will do faster AC WLAN than 400 MBit?

TP-link is not running in a CAPsMAN managed network. You can't compare managed network and simple AP situation. If you use hAP ac as a simple AP without any CAPsMAN, it will work at 400Mbit easily. This was mentioned above a few times already.

Will CAPsMAN support fastpath at some point?

Any Updates?

wAP ACs are actually available now. Would like to be able to use them where people are budgeted for UniF--K. But if I can expect 1/2 throughput... that's a problem.

Bump time.

Bump about what specifically? I can share results with more than 550Mbit iperf test through hAP ac over wireless, but it seems to me, a few posters here are talking about CAPsMAN. So question unclear

Bump about what specifically? I can share results with more than 550Mbit iperf test through hAP ac over wireless, but it seems to me, a few posters here are talking about CAPsMAN. So question unclear

Not only would I be very interested in these results, but also on how to recreate them. Got a Mac Book Pro here with 3 Streams and iperf ready to go! Never used Capsman but never got those speeds.

Would be interested to see this 500mbps throughput too, I'm running in AP bridge mode and even with a 3x3 adapter I still peak ~220mbps real throughput.

I've got one serving my home and an AC Lite serving my office. I came to this thread trying to see what the differences were between fp and cm2 on the AC hardware, I've got fastpath and fasttrack enabled, bridged, just passing from my desktop to my laptop. Bandwidth tests, file copy, download from my webserver (which is on the same switch interface but it is routed as it is a public IP vs my private IP), I get 340Mbps copying from my desktop to my wireless laptop, under 200Mbps copying to my desktop, 80MHz channel, 760Mbps air rate, one wall between, -48dB, 99% ccq. FP seems to be stuck at around 200Mbps (20MegaBytes per second on average) on file copies.

Hardwired it maxes my gigabit network, 100MegaBytes per second.

I actually get slightly better throughput with CM2 than FP.

Bump about what specifically? I can share results with more than 550Mbit iperf test through hAP ac over wireless, but it seems to me, a few posters here are talking about CAPsMAN. So question unclear

I see a couple things.

One question was about fastpath support for capsman in the future.

Another was getting better capsman performing hardware to compete with other managed wireless solutions out there.

Here is something we tested here. The wireless config is default (but in addition, wmm was enabled). The router was configured in bridge mode, test is done over wireless from Macbook Pro (3 chains) to a x86 machine connected to the hAP ac ethernet port. iperf command visible in the terminal window title.

Here is something we tested here. The wireless config is default (but in addition, wmm was enabled). The router was configured in bridge mode, test is done over wireless from Macbook Pro (3 chains) to a x86 machine connected to the hAP ac ethernet port. iperf command visible in the terminal window title.

Thanks for sharing, I never got values this high. What channel was that on? Also, if that iperf3 command is from the MacBookPro, then you seem to have tested Uploading. Could you please also share your -R (Download). In my tests those are usually (and oddly) lower than upload.

I just did the exact same test, Normis, got around 450Mbps with a dual chain client using CM2, about what I expected, maybe a hair lower than I'd like. Stuck at 200Mbps using FP. No other changes were made aside from enabling the other package. Fast track and fast path are enabled.

ROS 6.34.6

Bump about what specifically? I can share results with more than 550Mbit iperf test through hAP ac over wireless, but it seems to me, a few posters here are talking about CAPsMAN. So question unclear

I see a couple things.

One question was about fastpath support for capsman in the future.

Another was getting better capsman performing hardware to compete with other managed wireless solutions out there.

Normis?

Still wondering if there is any trick or config change I can do to get Caps-Man APs to pass better than 200 M?

Just spent several hours trouble shooting UniF--K. PAINFUL!

I have done a few Caps-Man systems now... I have to make absolutely sure that the client has 100M or less service.

with the latest firmware, i seem to be getting 250mbps through a wall or two, and it seems to go up to maybe 300mbps before hitting 100cpu on the wap ac.
no local forwarding, just client forwarding enabled. somehow local fwd is not stable at all for me (or wasn't on prev versions)

I seemed to have dropped down to about 130Mbps on average downloads.
6.39.1 on RB3011 and hAP AC.

Go to bridge, settings, uncheck use ip firewall.