AX6000 1 to 2 meters proximity ping from clients

[tashcz]

The 1G+ connectivity isn't used for WAN, it's used for my home LAN. When you transfer a couple of hundred gigabytes of files, it really makes a difference, especially when you're on WiFi. My other hobby is drone flying and I have a lot of 4k60fps footage that I backup to my NAS. Each time I go somewhere and do a few flights, it's at least 100-200GBs of footage. I carry my laptop mostly if I'm on a remote location (vacation, weekend etc) so and SD card sometimes isn't enough. When I get home and copy all the files to my NAS it makes a ton of a difference if it's 25-30MB/s to NAS or 120MB/s via wired. I can achieve 200MB/s with those drives. Also, I have backups of my Steam library that I pull from my NAS, games today are 100-150GB. It's not just about internet

I've pinged my devices from the router in most cases since Windows doesn't do sub milisecond response timings, and most of my clients are Windows, except the home servers that also had additional latency (around 1ms more) than on the AC58U.

I've done extensive tests over the last couple of days, not only ping tests, iperf, but recorded the loading of certain web pages via highspeed camera and compared frames it takes to do something. The AX6000 was about 50-250% worse than my AC58U.

Therefore I returned it. As it was adding latency and going all over the place even via wire, it must be either the routing function of it or the internal switch. I hope to get another AX6000 without those issues. If not, I'm really eager to switch to MikroTik. For this kind of money I really don't want a consumer router to do this kind of crap. I know it's not enterprise, it's SOHO, but I'd rather spend a few days setting up the MikroTik solution than debug what's wrong with the router for 2 weeks. And I know it's cheap for what gear like MikroTik costs. That's why we get those home all-in-ones

All in all, short overview:

Wired LAN: AC58U 0.5ms, AX6000 1-2ms
Wireless LAN: AC58U 1-3ms, AX6000 2-8ms average
WAN to 8.8.8.8: AC58U 12-13ms, AX6000 16-19ms avg.

Over subjective tests, Instagram image loading was on average 3-4x slower than on AC58U via frame counting. Youtube scrolling from bottoming out with gray images to loaded images took about 2x as much. Something is wrong. An AC58U shouldn't even be close to AX6000, let alone twice as better.

As far as ping results not being as extreme as percieved experience, I'd say there's a lot more going on when you actually load web pages, there's a lot more packages moving around and the latency impact is exponential.

Will post more once I get some info from the store.

 

[Volt]

A very strange situation indeed, I wonder whether another AX6000 will have the same behavior, please keep us updated. I was thinking about buying this router to somehow improve my Geforce NOW cloud gaming experience, but since my pings are not that good already, I probably should look for another model.

 

[L&LD]

Sounds like the unit you have may be defective. This is certainly not my experience with the GT-AX6000.

Try another example.

 

[tashcz]

@Volt,

When GPU prices were inflated, I had to buy a GeforceNow subscription for my girlfriend. From my experience, there's no way GFN will work well via wireless. If there's trees around in the game, they'll look awful. For that, really, you have to use wired. The latency to their servers is mostly ~10ms and you can't play FPS games by no means. We mostly played 7 Days to die, where you could get away with the latency as it was as VSync was turned on. Wireless was a different story... very bad experience. Video quality wise, also latency was a lot more noticable since even 2-3ms will make a difference for those real-time apps. Don't plan on using that service with any kind of wireless. It's not something I read, it's first hand experience. It was sad, we started playing a lot of games together on covid lockdowns, but once we switched to some more powerful games, that was the only solution, and I literally couldn't get a GPU that'd handle 1440p for under 800-1000EUR. It was a different world once the prices dropped enough to get a Radeon 6750XT for her. Service is awesome, but lacks a bunch of games and really requires a good connection on every end.

 

[tashcz]

Got a new AX6000. Set it up and testing it currently. So far it seems to be "better" but not as good as AC58U was on ping tests. Seems like the internal switch isn't among the best in these units. Don't understand why, but if I get the same "jitter" via cable or wireless, it HAS to be the switch. Not sure how 2.5G ports are wired on the unit. There shouldn't be any additional routing if I'm on the same subnet, so that kind of isolates it. I'll also give it a day or two for channels to clean up as I'm going 160MHz, to test and clean half the band fully if possible.

If I really get the same results, I'll try setting it up as AP on another router and see how that goes. But at least now I'm hitting 1ms occasionally on 5GHz, and that didn't happen earlier.

What's weird is I got a factory sealed unit - everything's packaged as it should except the router didn't have as many plastic peels on it as the old one did, which was kind of odd. Either way, seems fine but I found that weird - the SN is new, it's a new unit. I filed a complaint and the very next day it was accepted that the unit was indeed faulty - not sure if it's because I wrote like 20 bulletins of stuff or they really tested the same stuff I did.

Give me a day or two and I'll get back with what I've got. If it turns out it's the damn switch, I'll grab a MikroTik to use it as a router and use this as an AP. It's "cheap" for what it gives as an AP too at least.

 

[drinkingbird]

Got a new AX6000. Set it up and testing it currently. So far it seems to be "better" but not as good as AC58U was on ping tests. Seems like the internal switch isn't among the best in these units. Don't understand why, but if I get the same "jitter" via cable or wireless, it HAS to be the switch. Not sure how 2.5G ports are wired on the unit. There shouldn't be any additional routing if I'm on the same subnet, so that kind of isolates it. I'll also give it a day or two for channels to clean up as I'm going 160MHz, to test and clean half the band fully if possible.

If I really get the same results, I'll try setting it up as AP on another router and see how that goes. But at least now I'm hitting 1ms occasionally on 5GHz, and that didn't happen earlier.

What's weird is I got a factory sealed unit - everything's packaged as it should except the router didn't have as many plastic peels on it as the old one did, which was kind of odd. Either way, seems fine but I found that weird - the SN is new, it's a new unit. I filed a complaint and the very next day it was accepted that the unit was indeed faulty - not sure if it's because I wrote like 20 bulletins of stuff or they really tested the same stuff I did.

Give me a day or two and I'll get back with what I've got. If it turns out it's the damn switch, I'll grab a MikroTik to use it as a router and use this as an AP. It's "cheap" for what it gives as an AP too at least.

There will be some additional latency when you go from 2.5 to 1 or 1 to 2.5, but it should not be significant, less than 1msec I would think. There's no routing involved but different serialization delays. If your pings are more consistent now and not spiking way up, I would not expect your wireless to always be 1msec solid, 2-3 is not unusual depending on your environment.

But no, the switches in these units are not stellar by any means either.

 

[coxhaus]

I get 1ms or 2 ms ping times and I am going through a wireless AP, L3 switch, and to pfsense from 28 feet away from the AP. My wife is also running a podcast using some Apple product in a different network in the L3 switch with a different AP as I hear it.


Pinging 192.168.0.1 with 32 bytes of data:
Reply from 192.168.0.1: bytes=32 time=1ms TTL=64
Reply from 192.168.0.1: bytes=32 time=1ms TTL=64
Reply from 192.168.0.1: bytes=32 time=1ms TTL=64
Reply from 192.168.0.1: bytes=32 time=2ms TTL=64

Ping statistics for 192.168.0.1:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 1ms, Maximum = 2ms, Average = 1ms

 

[Crimliar]

My own pings from multiple wifi devices to the GT-AX6000 router can be very high - higher than I can ping some external sites (fast.com is 4ms while my internet gateway is 10ms)! But I can also ping AX to AX client or AX to LAN clients on the network at roughly 0.5ms.
Everything is working, well and fast. I have no need to be pinging the router, so I may not understand why it's working that way, but neither am I particularly bothered.

 

[tashcz]

Ugh, no sub-ms on my WiFi. Most clients are Windows 11, so I gotta enable ICMP responses individually. Wired I never get that 0.5ms, even pinging devices from the router itself. It's not my concern though (at least not the only one), the consistency is. I have a couple of servers in a DC wired up in LAN too, and that is like 100-200ns constant. I know that's enterprise networking though, don't expect the same results on a cheap (after all it's an all in one unit) home router. Anyway, I moved on from testing ping times, moving on to how it loads content on web and it's kind of weird too. Like, opening up a news portal, I'd get 10/20 images to load instantly and the rest to load up in 2 seconds (no lazy loading on websites and they're all visible). It's really hard to isolate what the problem is here.

Again to note, I'm probably asking for too much out of this, but something really feels off. I've played a few CS2 matches (FPS shooter, don't expect everyone here to be into gaming) and the latency is okay, there's no jitter detected. What I've found is that it's definitely not wireless related - wireless is only where it became most noticeable.

Anyway, let me attach this image from WiFi insights. Is this number of errors okay?

 

[sfx2000]

Just to throw something out there - Sustaining Pings to a Rpi 3+ (dual band) over WiFi...