Why is the AX86U considered better than the AX88U?

Spartan

Regular Contributor
Everyone here seems to recommend the ASUS RT-AX86U over the RT-AX88U. What is the reason behind that?

I have the RT-AX88U but it keeps disconnecting randomly if I use any firmware from thee 386.xx and above so I stuck to the 384.XX firmware (PS: I even tried Merlin's firmware several times but the random disconnections are still there as many other users on this forum have experienced).

what I've done is I purchased a Netgear AX6 router and set it up as the main router and set up the RT-AX88U as an extender solely to have my LT2P VPN configured where I've attached it to my IP Phone for work.

Also, what do you think of this RT-AX86U ZAKU II Edition? Is it the same but with a different design?

Do you think it is wise to keep my current setup with the Netgear AX6 as the main router or shall I get an AX86U and make it the main router and keep my AX88U for the VPN/IP Phone?
 

eibgrad

Part of the Furniture
Beyond the obvious of the 88U having four additional LAN ports, I don't see all that much difference between the two.


Although the presumably lesser of the two, the 86U, has the 2.5 gigabit WAN, and two WAN ports. So in some ways it would appear superior to the 88U. I think the additional LAN ports makes it compelling to some, but that comes at a slightly higher price (~$50 USD). Plus, it's pretty clear the 86U has become the "go to" router around these parts, at least for those using Merlin. I just think the combination of most bang for the buck, plus proven stability, has made it the de facto standard.

Seems a little late to making this decision if you've already committed to the Netgear AX6.
 

L&LD

Part of the Furniture

Tech9

Part of the Furniture
Also, what do you think of this RT-AX86U ZAKU II Edition?

The same router in red. Also available in some countries in yellow.

1656543520642.png

1656543297611.png


If you can wait, you may have a chance to get an updated RT-AX86U Pro:

 

SAL9K

Regular Contributor
Mere specs are not indicative of what to expect. Only comparison testing brings any/real differences to light.

RT-AX86U vs. RT-AX88U

Asus RT-AX86U review

386.1 Final 2x RT-AX86Us 2.5GbE Backhaul


The RT-AX86U has better WiFi RF (i.e. throughput) lower latency and the 2.5GbE Port advantage, including a newer SDK.

Not only is it cheaper than the RT-AX88U, it is the far better router too.
I’ve been having a hard time reconciling the SNB performance charts when comparing the ax86u/ax88u, which show that multi band latency (upload ping latency under max download) is far worse on ax86u??

The throughput is much higher on the ax86u due to 2.5GbE WAN used for the testing, but the latency is very poor...

https://www.smallnetbuilder.com/too...-asus_rtax86u/4731-asus_rtax88u?testmethod=44
 

Tech9

Part of the Furniture
I’ve been having a hard time reconciling the SNB performance charts when comparing the ax86u/ax88u

Testing was done in isolated chamber. This is not real use conditions. I had AX88U for experiments before and now I have AX86U for the same purpose. In my Wi-Fi environment AX88U was the better router with stronger signal. AX86U Wi-Fi is a bit of disappointment for me, honestly. Also, according to Wi-Fi Alliance AX88U is Wi-Fi 6 certified router. AX86U doesn't exist in their database.
 

SAL9K

Regular Contributor
The charts show that 2.4G performance is slightly better on the ax88u, but 5G is much better on the ax86u. The total throughout can’t really be compared because the ax86u has the 2.5GbE WAN, so it’s not even close there, ax86u ~50% higher.

However, the latency is much better on the ax88u, which is what doesn’t make sense to me, as you’d think that latency has a ~1/throughout inverse relationship.

A controlled testbed is the only way you can objectively quantify performance.
 

Tech9

Part of the Furniture
A controlled testbed is the only way you can objectively quantify performance.

I agree, the synthetic performance. I see the opposite though in my simple tests with the same clients at the same distance and behind the same walls. About latency - the two routers provide exactly the same user experience. There is no way to "see" few ms difference, even if the user is a hardcore gamer.
 

Tech Junky

Very Senior Member
2.5GbE WAN
Doesn't matter when the ISP link is under 1GE. The performance numbers on WIFI are typically more important for LAN traffic.

When you're trying to stream from your NAS then better WIFI is a key ingredient. The WAN port speed doesn't matter unless you're paying for those kind of speeds.
 

L&LD

Part of the Furniture
Trying to reconcile test scores in 'controlled' environments is easy. Test in your own home/network and believe the results you get there.

Unless you're maxing out the performance of the up/down link 100% 24/7/365, then the synthetic scores may matter to you.

In every other scenario in real-world use, I saw the opposite.

And, btw, I don't have greater than 1Gbps speeds over the WAN either (i.e. 'paid-for' speeds).
 
Last edited:

Tech Junky

Very Senior Member
Having done this personally to get more than the ~960mbps out of a gig subscription I used higher than or more than 2 gig ports to bundle in LCP through the modem to exceed the speeds of the plan. W/ 2 ports in LACP on a gig plan I was hitting anywhere from 1300-1500mbps. So, it's possible if you subscribe to a higher plan to get more bandwidth but, chances are you're using different gear and techniques to get these speeds.

When I switched to a gig plan years ago I switched back to an Asus for the gig ports since I swapped to D3.1 modem that was needed for the plan. The Asus fell on its face and couldn't achieve the speeds it was designed for on the WAN or LAN but hadn't been tested for those in a prior setup.

Put the device through the paces before the 30 day window closes is the lesson here. Relying on a POS you can buy at the corner store is just dumb.
 

SAL9K

Regular Contributor
I think the SNB is testing for the devices' maximum capability, and it'll most likely hold true that a more capable device (ax86u, better 5G radios, 2.5GbE) will perform about the same (or slightly better) when used at a lower user requirement (I.e. ISP <1Gbps), as a competing, lower spec device (ax88u). Have you read the testing setup, it's super legit, with really expensive RF testbed gear.

https://www.smallnetbuilder.com/wireless/wireless-howto/33224-how-we-test-wireless-revision-11

If you look at the 5G uplink/downlink test charts, it's clear the ax86u is like 20% better on 5G (no 2.5GbE involved here, it's just a better performing radio). That's the data. "Works better at my house" is not data, unfortunately.

However, I would like to have seen the same latency measurements made with the ax86u using its 1GbE WAN, as that would indeed have been a more fair, more typical user case.
 

SAL9K

Regular Contributor
I think the ax86u is clearly better if:
-- You're 5G client heavy
-- You have an ISP plan >= 1Gbps (w/ 2.5GbE capable modem)
-- You have wired backhaul to another 2.5GbE capable AP (E.g. another ax86u, or similar)

The ax88u is clearly better if:
-- You absolutely need 8x1GbE ports

Otherwise, they're both about the same.
 
Last edited:

L&LD

Part of the Furniture
My seat-of-the-pants real-world test of network latency is how fast I can click on a link and see the next word I need to read (yes, the internet is just a book to me).

The RT-AX88U was fast, like a souped-up dump truck. The RT-AX86U was more like a small/light sports car with an engine the same size as the RT-AX88U has. Much faster (lower latency) results were immediately noticeable (and still are on different customers' networks vs. mine).

They're not anywhere close.

The (surprise) baby brother, the RT-AX68U is almost at the same level (and much above any other AC class router I've used before).

As already mentioned, the newer SDK and/or RF improvements in the two years between the 'AX88U and the 'AX86U and the 'AX68U may be the direct reason it's so noticeable to me.

With the RT-AX88U, my network was fast and stable with the maximum speeds being contained to a corner of my home.

With the 2x RT-AX86Us in wired 2.5GbE backhaul mode, those maximum speeds are uniformly available 'everywhere', including to any wired devices I want to connect to the AiMesh node's location.

I don't believe you read too much in the links I provided above (please take the time to do so, if you haven't already), but no combination of routers could hold a candle to 2x RT-AX86Us in wired backhaul mode. Even when the RT-AX88U was added to the network in any fashion, the latency of the network was immediately and obviously worse.

For my LAN devices, after I added the QNAP 2.5GbE switches, the benefits for raw throughput were obviously obvious too (between wireless devices at 160MHz band width and/or wired (2.5GbE) devices to my (2.5GbE) NAS boxes.

I've had 1Gbps Fibre ISP speeds (symmetrical, up/down) for a few years now. So far, not even 2x GT-AX6000 routers match the performance of the 2x RT-AX86Us (and the 'specs' on the GT's are above the RT's, on paper).
 

SAL9K

Regular Contributor
I've had 1Gbps Fibre ISP speeds (symmetrical, up/down) for a few years now. So far, not even 2x GT-AX6000 routers match the performance of the 2x RT-AX86Us (and the 'specs' on the GT's are above the RT's, on paper).
Wow, ok, you're saying that not only is the ax86u better than the ax88u, it's better than the newer gt-ax6000. That's interesting.
 

L&LD

Part of the Furniture

SAL9K

Regular Contributor
My theory on the SNB latency charts, and why the ax88u scores better... the SNB "multi-band" tests are done from the sum of two clients (single 2.4G 40Mhz + single 5G 80MHz client) through to the WAN. The ax86u clearly is going to provide more "throughput" as it has the 2.5GbE WAN, and it does ~1.4Gbps versus the 940Mbs of the ax88u (limited by the 1GbE), which is 50% more total WAN-to-WLAN throughput.

However, for the multi-band latency tests, the same 2xclient loading is used, therefore the uplink ping latency (WLAN-to-WAN) is being measure with:
-- ax86u loaded to 1.4Gbps (WAN-to-WLAN)
-- ax88u loaded to 940Mbps (WAN-to-WLAN)

They have similar CPU's (quad core 1.8GHz), so the latency should be higher on the ax86u, as its client load is 50% greater during the testing. Incidentally, the ax86u latency is about ~46% higher (22msec versus 15msec), which is on par with the throughput difference. To be a fair comparison, the ax86u should have been tested with the 1GbE WAN. Even better would be a latency vs. throughput chart, I.e. this is how much latency to expect for "my client loading", or rather "for my max ISP speed", which would give the data to honestly compare across routers. SNB, please add that!

Additionally, now that I think about it, the ax86u is managing a 20msec ping time (on 5G) while clients are downloading at 1.4Gbps -- this is superb!

EDIT: Added link for reference:
https://www.smallnetbuilder.com/too...-asus_rtax86u/4731-asus_rtax88u?testmethod=44
 
Last edited:

L&LD

Part of the Furniture
You are analyzing things far too much.

I still don't believe you have read any of the links I've posted?

The lower latency and greater throughput I have seen are from the RT-AX86U itself. Regardless of which LAN port was used. From the same 1Gbps Fibre (symmetrical) ISP.

Most times I get around 2ms loaded and 0ms unloaded times. Now, near local peak times, I'm at 1ms unloaded and 4ms loaded.

Access to the NAS' GUI (whether it is connected to the 1GbE or the 2.5GbE ports (on each end) is faster than the RT-AX88U ever was. When actually transferring files, of course, the 2.5GbE port trumps the 1GbE port from the NAS to any other client device in the home.

The routers don't have just the main CPU, they have other helper CPU/SoCs too. Looking or calculating the load on the different routers based on the max throughput possible and the SoCs used is not at all accurate or the way to see which is the better router platform (overall).

I know what the differences are because I had and tested both in my home. I also tested many different variations of main/node setups, and with up to 6 different routers too. Two RT-AX88Us in wired backhaul mode were not worth it to me (my main use of such a setup, other than more even wireless? Wired connections at the other end of the house).

However, two RT-AX86Us were definitely worth it (sale prices, obvious throughput, and latency benefits, via wired/wireless), and the fact that I had pre-sold the RT-AX88U before deciding to keep a 2x RT-AX86U setup in my home.

I didn't buy those two routers because it was the cheaper choice.

I didn't buy to just have the 'latest and greatest' model.

I didn't buy to simply add a wired network connection at the other end of the home (a $10 switch could have done that).

I bought them because, with the same ISP (1Gbps up/down) speeds, my network was not only faster. But it had consistently better coverage (of course, 2 routers should do that... yet, some don't). And stability and reliability that have proved to have exceeded the RT-AX88U I had so long ago.

All of the above holds true when compared to the same setup with 2x GT-AX6000s. A single GT-AX6000 is superior to the RT-AX86U. More throughput to the same client/distance, effectively the same latency as the RT-AX86U. When a second wired (2.5GbE) backhaul GT-AX6000 (and/or one of my RT-AX86Us), the reliability, stability, and consistency of the network fell apart.

No charts are telling me this (nor will you find this anywhere from a 'review'). Comparison testing did.

You may not have the same results in your network and WiFi environment.

But without testing, you will never know (from just reading the web).
 

SAL9K

Regular Contributor
I did read your posts and reviewed your links; I stopped at the CNET article though. Also, your latency 1msec/5msec Fast.com results must be over cable. The SNB (Small Net Builder, @thiggins) multi-band latency testing is WAN-to-WLAN (to wireless clients), w/ uplink ping latency, which is representative of how responsive a client's wifi packets may be under max multi-band routing load. Your single cabled client Fast.com results are not a fully loaded router (with radio activity), as you'd need a >>1Gbps ISP plan *and* be using wireless client. The SNB testbed can drive the WAN to 10GbE, I believe, so the testing is truly router limited, not ISP limited.
 
Last edited:

Latest threads

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top