as the title goes. what's real wifi transfer speed of that router, hap ac? let's assume ideal conditions (no background noise, client in close proximity, updated fw, ac-only setting, Ceee wide channel...) and one ideal ac 3x3 client, which is not another mt device (meaning no cheating by means of nstreme or nv2).
and what's possible to get out of say 2x2 ac client, if someone knows.
i'm getting 260mbit between my killer 1535 wifi card and hap ac, cpu load on mt is about 30%, tx ccq about 50%. i think that's too little but i don't know really.

tnx.

50% ccq is really bad, for good connection you need more then 90%

yes.... any tips to improve it? on low speeds it's 100% (surprise, surprise) and my client connects with 866mbps, so i have chances at least in theory. it's my first mt device and i really like it, but to be honest, learning curve is a bitc* and documentation is really scattered.

How close to hAP ac is client? Too close (e.g. less than say 2 metres without any obstacles in between) degrades service as well.

mt at one end of the hall, attached on the brick wall at 2.2m height, which is 20cm under ceiling, upside down (wired interfaces facing up).
client was positioned in my hand from 1 to 8m away with no change in ccq, apart from when i was walking away or towards the router. signal should be about -50, -60db, correct?

i saw a post or a mum saying i could measure signal on each antenna to see if there is a large difference. is that possible within ros?

quoting myself here, but i just found the answer, it's in the wireless->registration->client i want to check. so, signal per chain strength is -50, -53, -50

edit:
i'm new also to the real world. before my mt i did only internet research and i underestimated the meaning of "perfect conditions" . so, i re-did measurements. distance between mt and client was ideal according to another topic on this forum - signal strength ~ -50db, floor noise -100db.
results: spikes of 400mbps down, 500mbps up, average not much under it. (hehe i actually had to pick bigger file to transfer). not sure why uplink is faster, but that's life.

now the big question: i don't like standing in the middle of my hall, 3.4m away from mikrotik, to get good wifi. is there anything i can do to improve reception out of ideal conditions? i read increasing tx power sometimes works, but mostly only over-stresses the transciever chip... i also don't want to move mt under my couch.

see viewtopic.php?f=7&t=132384

is there anything i can do to improve reception out of ideal conditions? i read increasing tx power sometimes works, but mostly only over-stresses the transciever chip... i also don't want to move mt under my couch.

When using 5GHz band (and higher), LOS gets quite important. Brick wall penetration is almost zero, so is refraction on corners. So ... it is important to have AP placed at strategic place, ideally close to point where clients mostly use it. Sometimes this means pulling ethernet cable through a wall or two.
To improve WAF (wife acceptance factor) I put AP inside some closet. Plywood absorption of WiFi signal is bearable.

yes i saw... moving client for 20cm so that it hides behind a corner of a brick wall halves the transfer speed instantly.
my urge to have 500mbps which i almost never use is not strong enough to drill holes

@eddieb, the 500mbps link you mentioned you achieved - what was the environment? I guess LOS, snr > 50db, -55db on all 3 chains, about 3-8m away from router?

When you have an urge to use WiFi around a corner it's better to use 2.4GHz band (if you don't use it already). That highly depends on interference from neighbours' APs of course.
My RB951G, hidden inside a closet, behind a brick wall (or around two corners if that signal actually travels through the door), up the stairs ... and still at -75dBm, enough to do some web browsing, youtube movie watching, etc. 5GHz is long gone by there (did a test with RBD52G, which is dual-band. 2.4GHz was similar to just described).

depends on the corner, really. and still half of 400mbps is 200mbps, which is still twice as what i can get over crowded 2.4g space.

Real world result from a phone in a room across from the hAP AC (wall mounted high up). Almost clear LOS (has to go through a doorway). -57 dBm on the hAP AC, -54 dBm on the phone. Upload limited by ISP.

With 3 chain client (like macbook pro) we have seen above 500Mbit iperf3 results.

Shouldn't it be much better?

no. as it has been discovered, CPU gets capped at 500mbps and that's the end of it. that speed is reachable even with using 2-chain client, such as on my dell xps.
now i only need to figure out why i can only reach that upload, while download stays below 400mbps. could it be my 2 little receiving chains?

When testing my hAP ac2 I noticed that CPU load is higher during DL than during UL which made my ac2 faster in UL than in DL.

not in my case. 350mbps has eaten about 60% of my CPU, which makes it *roughly* proportional to 500mbps being 100%

We did reach nearly 700Mbit, but that was inside an anechoic chamber with no outside interference, and with specific iperf3 settings. With normal conditions you should expect 500-600Mbit on 3 chains.

mind sharing your settings? I'll see if I can lower my CPU consumption. I did my tests by copying large file between NAT on wired IF and my wifi client.
ps. i've seen somewhere a "best 2018" test and the winner really was something, AC3xxx-something, 8-legged, 300€+ beast. First router to hit a 600mbps mark. Normis, don't let them know.

No special settings. Wlan1 in bridge with Ethernet

No special settings. Wlan1 in bridge with Ethernet

Image Pasted at 2018-10-2 12-01.png

RouterOS 6.99 in 2017? Release 7 is near

hehehe what's been leaked cannot be un-leaked:)

one last thing - what affects ccq? meaning, i have good ccq (>90%) while sending wifi towards mt, but terrible (~60%) when receiving. same spot, empty 5g band space etc. mt says -50dB on all chains, and i measured about the same on client side.

There always was 6.99, even years ago. It is special testing version of routeros and has no relation to v7. I also had a router with this version in the past...

No special settings. Wlan1 in bridge with Ethernet

Image Pasted at 2018-10-2 12-01.png

Nice result. But now the same test with 6.43.2. Thanks

No special settings. Wlan1 in bridge with Ethernet

Image Pasted at 2018-10-2 12-01.png

Nice result. But now the same test with 6.43.2. Thanks

Maybe it is ROS with origin ATH10k?

Hate to nag our respected elder speed fanatics, but does anyone have a suggestion about my asymmetrically crippled ccq? Is it possible a bad mt wifi transmitter and in that case how would i check it?

Did you fiddle with tx power settings on routerboard?

I was ready to punch in some number for manual power output, but then I read it's bad, so I didn't
Anyhow, the manual is also kinda vague about what that number means - is it some kind of coefficient, factor... https://wiki.mikrotik.com/wiki/Manual:W ... he_best.3F
what's the "argument passed to eeprom function"? percentage?

okay, but i can try to fiddle with power to lower it. perhaps lowering output power is a way to see if the transmitter works sub-optimally at its max power. What is the number I start with? 17dB, then down?

As a comparison yesterday I had a Macbook Air (v10.13.6) download OSX 10.14 from a local Mac running a cache at over 40MB/s from a WAP-AC (v6.43.2) a few metres away. I'll try again in the next few days with a CAP-AC and report here. This is on a CAPSMAN system with local forwarding. That's probably one of the best speeds I have ever seen from a Mikrotik and I was a bit surprised to say the least for a dual chain setup (Air is 2 chain, not 3 chain like the Pro's).

Makes you a better customer than me, easier to satisfy . I don't think CAPsMAN has anything to do with speed, it's just management system for when you have a grid WiFi to maintain. I have only 40MB/s and still ranting due to bad CCQ, which cure i'm searching for. I have 50MB/s on upload, cause CCQ is nice in that direction.
Anyhow, we're eager for u to share results!