[Asus RT-AX88U] Experiences & Discussion

[PhantomRay]

My RT-AX88U does seem to have a slightly better range than my RT-AC88U, the signal on my Zenpad is a bit stronger while at the other end of my appartment.

Yes, comparing to AC5300, AX11000 did have better range or better signal strength for the devices on further distance on a different floor. However, when clsoe range, same distance, the AC has stronger signal 30-40 dB from AC vs 40-50 dB from AX. I got the reading from my PC's network analyzer tool though. I haven't check out the AX88u.

But all AX routers I had tried have better 2.4Ghz range, that is for sure. And I think AX88U did better than my AX11000, maybe because of a newer firmware?

 

[frichardson]

@PhantomRay - Very interesting reported number for the GT-AC5300. I was very curious about getting higher than the ~640Mbps d/l speeds I'm seeing now and I realized two things: (a) my Laptop has an Intel AC-8026 dual band adapter with a maximum theoretical limit of 867Mbps and (b) the peak d/l speed of 941Mbps in the review you link above is running with a 4x4 AC device which the RT-AX88U should also support (but I unfortunately don't have one...), the other numbers in that article are on par with what I'm seeing (~600Mbps at low attenuation).

The article here on the RT-AX88U also gives similar 80MHz AC throughput (around 600Mbps at low attenuation):
https://www.smallnetbuilder.com/wir...-first-wifi6-peek-asus-rt-ax88u-netgear-rax80

Instead of 4x4 performance numbers, they get into 160MHz performance which if I updated my adapter to a AC 9260 I might be able to take advantage - but the low attenuation performance for that adapter is also around 850Mbps so it's not super exciting.

EDIT: actually the AC-9260 is only about $20 on Amazon - that might be worth a try...

 

[frichardson]

Well, I am pretty happy with the RT-AX88U so far. I did order a 160MHz adapter for my laptop - I'll report back on that.

@RMerlin - I thought I'd hold off on trying out your firmware for the next release, but if there is particular testing (beta or otherwise) you need or want done let me know.

 

[frichardson]

So I upgraded my laptops network adapter from the Intel AC-8260 to the AC-9260 which essentially gives you a 160MHz client. I enabled 160MHz on the RT-AX88U and the connection is shown both on the router and on my PC as supporting 1733 rx/tx. Speedtest.net gives about the same performance as before, but I don't know if the weak link is the ISP or not.

I guess I really need a way of measuing the rx/tx speed to and from the router without going over the internet. Would I do that by attaching a disk drive to the router and measuring file copy speed?

EDIT: okay, I clearly don't know what I'm doing. Changing the server on speedtest.net made a big difference. With the 160MHz connection I got 911Mbps which is quite a lot faster than what I was getting with the 80MHz connection (with the same server). But it's pretty clear that the only meaningful measure is to/from the router since the ISP can be all over the map.

 

[RMerlin]

I guess I really need a way of measuing the rx/tx speed to and from the router without going over the internet. Would I do that by attaching a disk drive to the router and measuring file copy speed?

Best way is to run iperf on two computers on your LAN (one on Ethernet, and another on your wifi client). You could also run iperf on the router itself, but the bottleneck might become the router's CPU then.

If you prefer a simpler GUI-based test, look at LANBench (https://download.cnet.com/LANBench/3000-2085_4-10974206.html), which I've used in the past.

 

[frichardson]

Thank you @RMerlin ! I saw that you have iperf included in asuswrt, but it didn't occur to me that the router might not be powerful enough to do this kind of benchmark well. Ethernet to wifi makes a lot of sense and I'll give that a try.

 

[Adamm]

Did some more benchmarking today with my new QNAP TVS-672XT, wasn't able to break 1.2Gbps in real world tests but had fun trying

 

[RMerlin]

Did some more benchmarking today with my new QNAP TVS-672XT, wasn't able to break 1.2Gbps in real world tests but had fun trying

And that's with a signal quality of around 50%? Surprising. Try getting closer to see how fast you can push it

 

[Adamm]

And that's with a signal quality of around 50%? Surprising. Try getting closer to see how fast you can push it

I think that's just a visual bug with the Intel 9560 driver any time it connects at 1.7Gbps. With that being said I was able to maintain this speed across my apartment, but we seem to bottleneck at this point though with a two stream client.

 

[L&LD]

Did some more benchmarking today with my new QNAP TVS-672XT, wasn't able to break 1.2Gbps in real world tests but had fun trying

Can you give a quick overview of how your QNAP is configured?

Btw, with a current WiFi backend, a few more LAN ports added and a router 'os' installed, this platform would make my dream router a reality. QNAP, what are you waiting for?

 

[Adamm]

an you give a quick overview of how your QNAP is configured?

Unfortunately because the AX88U only has gigabit ports I had to get creative... I have two Ethernet connections in LACP (192.168.1.69), then a one standalone gigabit link. (192.168.1.70)

Then to prevent a bottleneck I pulled one file off my m.2 read cache and another off a separate SSD.

 

[frichardson]

Oh right... this is actually pretty funny. The good part is that the 160MHz connections get me to the point where I can max my D/L capacity of my ISP - that is 1Gig and I get about 930Mbps. But, since the ether ports on the RT-AX88U are all 1Gig, I can't really go faster than that via iperf and another PC wired to the router. So, is there a chance with this slightly beefier Arm processor on the router that it could handle iperf? It seems like counting packet throughput shouldn't be a very hard task...

 

[Adamm]

Oh right... this is actually pretty funny. The good part is that the 160MHz connections get me to the point where I can max my D/L capacity of my ISP - that is 1Gig and I get about 930Mbps. But, since the ether ports on the RT-AX88U are all 1Gig, I can't really go faster than that via iperf and another PC wired to the router. So, is there a chance with this slightly beefier Arm processor on the router that it could handle iperf? It seems like counting packet throughput shouldn't be a very hard task...

You can download iperf via entware when on Merlins firmware, it works fine in my experience.

 

[frichardson]

You can download iperf via entware when on Merlins firmware, it works fine in my experience.

Thank you! I'll try that out - I'm really looking forward to punting the stock FW and install @RMerlin's magic

Do you happen to know what actual throughput people have seen with the 160MHz AC connection? If I understand correctly, 1700Mbps is a theoretical (calculated) bandwidth...

Sent from my Pixel 2 XL using Tapatalk

 

[Adamm]

Do you happen to know what actual throughput people have seen with the 160MHz AC connection? If I understand correctly, 1700Mbps is a theoretical (calculated) bandwidth...

Look up 5 posts, I posted a benchmark yesterday.

 

[Grisu]

Do you happen to know what actual throughput people have seen with the 160MHz AC connection? If I understand correctly, 1700Mbps is a theoretical (calculated) bandwidth...

1700 is max. connection speed (with best WiFi connectivity nearby), as rule of thumb you can say 1/3 is overhaed and 2/3 real throuput.

 

[frichardson]

Look up 5 posts, I posted a benchmark yesterday.

Yes, I saw that, but I was wondering if what you were getting is typically (i.e. if anyone ever gets the 1700Mbps).

1700 is max. connection speed (with best WiFi connectivity nearby), as rule of thumb you can say 1/3 is overhaed and 2/3 real throuput.

That's around what I was seeing with the 80Mhz AC (650 out of 850 or something). I noticed the Intel docs say "theoretical max"...

 

[frichardson]

In the notes I've read here, people seem to recommend avoiding DFS channels. I ask because for 5GHz, my neighbors are on top of the non-DFS channels. This definitely affected my throughput - a lot. Switching to a DFS channel seems to have fixed this, but am I setting myself up for other problems?

 

[L&LD]

Nobody will ever get 1700Mbps throughput. The overhead of WiFi ensures that.

If you're not currently in a restricted zone (planes, radar, etc.) using the DFS channels will seem great, because others will be avoiding them by default. I don't think it is legal to do so though? Something to consider as everything, including our air space, gets assigned and redefined continuously, depending on who is in power at the time.

 

[frichardson]

Nobody will ever get 1700Mbps throughput. The overhead of WiFi ensures that.

If you're not currently in a restricted zone (planes, radar, etc.) using the DFS channels will seem great, because others will be avoiding them by default. I don't think it is legal to do so though? Something to consider as everything, including our air space, gets assigned and redefined continuously, depending on who is in power at the time.

I think in the US the FCC re-enabled some DFS channels for use (https://en.wikipedia.org/wiki/List_of_WLAN_channels), but right I shouldn't assume that just because my router lets me do it it's okay...

EDIT: and right, I was just wondering how much over 1,000Mbps I can get with the 160MHz AC adapter. More as a geeky thing to figure out lol.