[HN Gopher] From 400 Mbps to 1.7 Gbps: A WiFi 7 Debugging Journey
___________________________________________________________________
From 400 Mbps to 1.7 Gbps: A WiFi 7 Debugging Journey
Author : tymscar
Score : 153 points
Date : 2025-11-01 19:50 UTC (1 days ago)
(HTM) web link (blog.tymscar.com)
(TXT) w3m dump (blog.tymscar.com)
| tripdout wrote:
| > Running iperf server on the router itself creates CPU
| contention between the WiFi scheduling and the iperf process. The
| router's TCP stack isn't tuned for this either. Classic mistake.
|
| Can you elaborate on this? I don't know much about WiFi so I'm
| curious what CPU work the router needs to do and what wouldn't be
| offloaded to hardware somehow (like most routing/forwarding/QoS
| duties can be).
| Nextgrid wrote:
| It has nothing to do with WiFi even; when running a test you
| need a server that emits the test data - this could be a
| standard HTTP server on the internet (in case of public speed
| tests) or a binary like iperf that synthesizes the test data on
| the fly.
|
| You need to ensure the server is able to send the test data
| quickly enough so that the network link becomes the bottleneck.
|
| In his case he was running the test server on the router, and
| the router's CPU was unable to churn out the data quickly
| enough to actually saturate the network link (most network
| equipment does the network switching/routing/NAT in hardware
| and so doesn't actually come equipped with a CPU that is
| capable of line-rate TCP because it's not actually needed in
| normal operation).
| mattlondon wrote:
| I had a similar issue but on unifi gateway lite after upgrading
| to 1gig fibre, I couldn't get above about 250-300mbps, even
| wired. Everything looked good in the unifi _app_. Turns out in
| the unifi web UI there was a "use hardware acceleration"
| checkbox for the gateway that was unticked and not even visible
| in the app. Ticked that and now I am getting 900+mbps
|
| I also sometimes have alerts saying more than one device is using
| the same IP address (DHCP issues) but it won't tell me which
| ones! At least give me the MAC addresses!
|
| Unifi's stuff is great, but the software is sometimes
| infuriating.
| scrps wrote:
| Other trap is some of the unifi features, IIRC their IDS is one
| of them, will cut throughput if you are running it.
| tymscar wrote:
| You're right, however, that was one of the reasons I upgraded
| too. This one can handle the full 2.5 gigs even with IDS on.
| nyarlathotep_ wrote:
| What hardware are you using? I'm not seeing near my
| advertised (and previously achieved, via Acer 'Gamer
| Router') with IDS on.
|
| IDS is probably overkill for a home network anyway.
|
| I recently replaced said router with a Dream Router 7.
| tymscar wrote:
| The maximum routing speed Unifi Dream Router 7 can do
| with IDS on is 2.3Gbps according to their spec sheet
| scrps wrote:
| Ah haven't looked at their current offerings in a while, I
| am still on the first gen USG 1000fdx but hand-rolling a
| 2.5 router (Radxa E52C, it's nifty) to replace it when I
| stop being lazy.
|
| You are right about Unifi's software being pain and I love
| that they keep changing the UI, the controller on the
| server side is dependency hell, and mongodb to boot just in
| case you need to manage n^webscale deployments.
| rconti wrote:
| I'm lazy so I just fire off the occasional speed tests using
| Ookla.
|
| It doesn't _really_ seem to matter what channel width or
| frequency I use, I tend to get around 600Gbps from my iPhone (17,
| pro).
|
| When you make it a point to ensure you're on the correct AP, line
| of sight from a few feet away, sometimes I break 1Gbps. I was
| surprised, watching TV the other day, to randomly get a 1.2Gbps
| speedtest which is one of the faster ones I've seen on WiFi.
|
| (10gbps internet, UDM Pro, UDM enterprise 2.5Gbps switch for
| clients, PoE WiFi 7 APs on 6ghz).
|
| Honestly, I'd say overall 6ghz has been more trouble than it's
| worth. Flipping the switch to WPA2/3 as required by 6ghz broke
| _all_ of my clients last year, so I had to revert and now I just
| have a separate SSID for clients I have the energy to manually
| retype the password into. 6Ghz pretty much only works line of
| sight and from a handful of feet away. There were bugs last year
| in Apple's "Disable 6e" setting so it kept re-enabling itself.
| MLO was bad, so it would stick to 6ghz even when there was
| basically no usable signal.
|
| Over the course of the past year, it's gotten pretty tolerable,
| but sometimes I still wonder why I bother-- I'm pretty sure my
| real world performance would be better if I just turned 6ghz off
| again.
| manquer wrote:
| >> get around 600Gbps from my iPhone 17
|
| !
|
| What kind of magic iPhone you have? I don't think there is any
| device to achieve anything close to that today[1]
|
| ---
|
| [1] The recently(2024) record is claimed to be at 938 Gbps but
| it is only to a 12cm distance[2]
|
| [2]
| https://discovery.ucl.ac.uk/id/eprint/10196331/1/938nbspGb_s...
| vardump wrote:
| Obviously he/she meant 600 Mbps.
| TexanFeller wrote:
| I just tested 1700mbits/s from my iPhone 17 PM in the next room
| over from my Ubiquiti E7 and I don't even have MLO enabled.
| Something's very wrong if you're only getting 600mbit.
| ukd1 wrote:
| I get consistently ~1.3-1.6gbps on fast.com with similar setup
| (10g fiber, UDM Pro, E7, etc). I think where I live there are
| very few / zero folks on 6ghz...so, win.
| freetime2 wrote:
| I get 1,700 Mbps on Ookla with my iPhone 17 Pro. This is on
| 6ghz with line of sight to the AP, with MLO turned off.
|
| I haven't experienced any issues with 6ghz enabled, although
| honestly there isn't much noticeable benefit on an iPhone
| either in real-world usage. MLO was causing some issues for my
| non-WiFi 7 Apple devices - since WiFi credentials are sync'd in
| iCloud, I found that my laptop was joining the MLO network even
| though I never explicitly told it to - so I have disabled MLO.
| Nextgrid wrote:
| Optimizing for top speeds is the wrong way of looking at this.
|
| Even the shittiest consumer WiFi will generally give a
| satisfactory speed test result with decent speeds, despite
| being completely unusable for anything real-time like video
| conferencing, Remote Desktop or gaming. Your random high-speed
| result may very well be down to luck and doesn't represent how
| stable and usable the connection will be.
|
| In fact what the author does here (crank up the channel width,
| etc) might do for a good speed test result but will start
| dropping out with terrible latency spikes and jitter the second
| he turns away from his WiFi AP.
|
| Smaller channel widths are generally preferable as they provide
| a smaller top speed but said speed will be much more stable.
| rconti wrote:
| Sure. Those are also things I optimize for. I'm using 40mhz
| 5ghz channels and 20mhz 2.4ghz channels. I'm in the 'burbs,
| but silicon valley, and small lots, so there's definitely
| some contention for channels. Just sharing my experience.
| Avamander wrote:
| My MSM560 that's approximately 15 years old can do >700Mbps
| with a 13 Pro. If you're getting less on newer hardware
| something is terribly wrong.
| throwaway314155 wrote:
| > Flipping the switch to WPA2/3 as required by 6ghz broke _all_
| of my clients last year
|
| All? Really?
|
| > and now I just have a separate SSID for clients I have the
| energy to manually retype the password into
|
| Type it once and it will be saved, as has been the case for
| years.
| rconti wrote:
| Yep! Every single client required the password be typed in
| again, which is problematic in a house full of wifi devices
| (~50), some of which don't have keyboards, or have janky
| setup processes. Surely you're aware of wifi devices that
| need you to connect to their own SSID to set them up, or
| require an app and a setup process.
| vlan0 wrote:
| That's interesting. My testing for EAP-TLS and OWE networks
| has shown modern clients will simply create another profile
| when it detects the change in the AKM suite. Hard roam
| between wpa2/wpa3, but still seem less for the client.
| ashirviskas wrote:
| Huh, I have a random 2.5G Wi-Fi 6 router with 2.5G provider
| connection.
|
| I just tested 1.3Gbps through some reinforced concrete on Wi-Fi
| 6, no line of sight.
|
| Is all that tinkering really needed?
| lazide wrote:
| From how far away?
| eqvinox wrote:
| > Set transmit power to High
|
| Do NOT do this if you live in a densely populated area (e.g.
| apartment complex). You'll create noise for yourself and
| everybody else. Classic prisoner's dilemma - a few people could
| be assholes and profit from it, but if everyone's an asshole
| everybody suffers.
|
| General rule on TX power: start on low and increase only if you
| know (or can confirm) it helps. Go back down if it doesn't.
| neilalexander wrote:
| This may not help if you can't control your environment. You
| will often benefit from nearby routers hearing you and each
| other if you are forced to share a channel with them, as that
| is what enables the carrier sensing to work correctly.
| Otherwise neighbouring APs that can't hear your quieter use of
| the channel may shout over your devices rather than backing
| off, creating collisions and resulting in retransmits.
| eqvinox wrote:
| You're describing the situation where the prisoner's dilemma
| has already gone wrong, with someone else not-nice shouting
| over you trying to be nice.
|
| In other words: you don't need carrier sensing to work if
| you're not getting drowned in noise to begin with.
| Retric wrote:
| You can get this kind of interfere even if the signal from
| the router to your device is sitting just above the noise
| floor and the property next to you is doing the same thing.
| Both signals are so weak they get drowned out by an even
| weaker signal. The router on the other hand can't tell the
| message is corrupted until your device responds.
| appreciatorBus wrote:
| If you're using the same channel as a neighbouring router
| that's close enough to overpower yours then you've already
| lost, pick a different channel. If you stick to 20 mhz there
| are plenty options, even more if you are able to use DFS
| channels.
| martinald wrote:
| Wifi7 can use 320MHz channels on 6GHz. There's only 1 of
| those in many locations.
| appreciatorBus wrote:
| Yes, exactly, this means you shouldn't use 320Mhz.
|
| Find quietest 20mhz available on 5 or 6 GHz. It'll be far
| more reliable than trying to battle someone over the 320.
| tymscar wrote:
| But also faaaaaaar slower
| wpm wrote:
| How likely am I to even detect my neighbors 6GHz network?
|
| I live in a very dense part of Chicago. 2.4 and 5 are a
| minefield, just a thick soup of interference on
| everything but the DFS channels (which I get kicked off
| of too often being close to two airports). While it could
| be that zero neighbors have 6E or 7 equipment, I find
| that hard to believe, but nothing comes up on the scan.
| Brian_K_White wrote:
| You control your environment by not adding yourself to the
| dicks creating the bad environment. Everything else is just
| rationalizing for your own maximum convenience.
|
| There is no such problem as "you have to shout enough so the
| others hear that you're there". There's no such thing, by at
| least 2 different vectors. 1, They hear everyone just fine,
| weak and strong, all at the same time. 2, It doesn't matter
| even if they didn't, because you obviously hear them if
| you're getting clobbered by them, and so your router can
| channel hop around them even if they don't channel hop around
| you.
| arghwhat wrote:
| While indeed you shouldn't fix noise by shouting louder,
| your justification isn't quite right.
|
| 1. It's the AP that has to decide to change channel, and if
| you live somewhere with channel contention, from its
| perspective all channels will be busy. At that point, if
| your channel appears the _quietest_ (either by being the
| least noisy or by your clients not being active), then the
| AP will decide to clobber your channel. Their WiFi devices
| may also not hear you and won 't back off to give your
| airtime, even though you hear theirs and give them airtime.
|
| 2. Having your AP change channel (note: channel hopping is
| something else entirely, which isn't used for WiFi)
| wouldn't help when all channels are busy. As long as _your_
| usage appears quiet, _other_ APs will keep moving on top of
| you during their channel optimization.
|
| For residential, the only solution is to use technology
| that cannot propagate to neighbors. 5/6GHz and many APs,
| and good thick walls (mmm, reinforced concrete). WiFi
| channels is a solution to make a few bits of equipment
| coexist in the same space, but is of limited use when it
| comes to segregating your space from that of your
| neighbors. Especially if you want good performance, as
| there's very few wide channels available.
| lazide wrote:
| Notably even drywall attenuates 5/6ghz to an obvious
| degree. It's quite useful in apartments.
| russelg wrote:
| I'm the only person with a router that's broadcasting 6GHz in
| my apartment complex, so until that changes I'm gonna keep
| using High transmit power :)
| gerdesj wrote:
| Quite right. However, if your wifi bridge has an option for
| auto tuning the power then that might be a future proofing
| option, assuming that everyone uses it, which they probably
| won't _sigh_
|
| If wifi becomes a pain within a shared building then
| seriously consider ethernet. Slimline stick on trunking will
| hide the wires at about PS1-2/m. A box of CAT6, solid core is
| less than PS1/m. You will also need some back boxes, modules
| and face plates (~PS2.50 each) and a POST tool (fiver?) Or
| you can try and bodge RJ45 plugs onto the solid core CAT6 -
| please don't unless you really know what you are doing: it
| looks messy and is seriously prone to weird failures.
| lazide wrote:
| 'High power' on a router is mostly useful to borderline
| clients. Without that, they likely won't even be able to
| see it. It's hard to auto detect that situation initially,
| since how can you tell someone you can't hear to get
| louder?
| Aurornis wrote:
| There might be more people than you, but 6GHz with wide
| channels doesn't penetrate very far. You wouldn't be able to
| see all of the networks, just maybe your adjacent neighbors.
| andix wrote:
| In my experience concrete walls and ceilings in apartment
| complexes completely block 5 GHz signals. Even through modern
| triple glass windows most of signal is lost. I can't receive
| any other 5 GHz networks inside my apartment, but around 50 on
| 2.4 GHz, which makes 2.4 nearly unusable anyway.
| Aurornis wrote:
| This is even more true for the WiFi 7 frequencies at 6GHz
|
| The old tales about interfering with your neighbors,
| prisoners dilemmas, and claiming moral high ground from
| setting it to low is old school WiFi mythology that continues
| to be parroted around
| lotsofpulp wrote:
| In the US, I would venture at least half, if not more,
| apartment complexes have wood and drywall walls and ceilings.
| No concrete is used above the first floor.
|
| https://en.wikipedia.org/wiki/5-over-1
| Aurornis wrote:
| For the 6GHz frequencies used, this isn't really as big of a
| deal as everyone has made it out to be. The advice was shared
| in the early days of 2.4GHz WiFi with into 3 non-overlapping
| channels, higher penetration of 2.4GHz signals, and competition
| with all of the other cheap devices in the 2.4GHz space.
|
| The 6GHz space isn't even competing with classic WiFi. It's
| really fine. There's no prisoner's dilemma or some moral high
| ground from setting it to low. It will make virtually no
| difference for your neighbors.
|
| The real world difference is actually pretty minimal between
| power settings.
|
| The actual risk with modern hardware is that the high power
| setting starts running the power amplifier in a higher
| distortion area of the curve which degrades signal quality in
| exchange for incrementally longer range.
| myself248 wrote:
| Also, the higher frequencies are much more affected by
| absorption from little things like "walls" and "trees" which
| are occasionally part of the RF environment, so you're far
| less likely to interfere with your neighbors doing this, than
| you were with 2.4GHz.
|
| Also the reason it makes such an enormous difference to put
| your AP in the same room, if at all possible. Sneak a cable
| somewhere, park the AP in the far _corner_ of the room, sure.
| But with zero walls in between, it 's huge.
| nixpulvis wrote:
| I love that we're making wifi so high frequency that we're
| back to running cables to every room.
| tymscar wrote:
| Running ethernet to every room is always going to be a
| good idea
| lazide wrote:
| Or conduit so you can run something else too later if you
| want. There is a reason why commercial almost always does
| that. But also because they have money.
| noja wrote:
| > General rule on TX power: start on low and increase only if
| you know (or can confirm) it helps. Go back down if it doesn't.
|
| The people reading this are techies. Nobody else will do this.
| Either it should be built into the protocol, or the advice
| should be abandoned.
| varenc wrote:
| Agree you want TX power as low as you can, but in practice,
| I've always found there's at least once device in my house
| that'll benefit from an increased TX power. Also I generally
| just the FCC in setting reasonable power limits for what 'high'
| should be.
| loloquwowndueo wrote:
| Why do people need 2.5Gbps internet access or 1.7 Gbps on a home
| wifi network? What are folks doing at home?!?
| vardump wrote:
| To transfer files? Like large virtual machines, huge video
| files. Backup their files quickly. To support a homelab to
| learn new skills. To stream uncompressed video. To download 300
| GB monster games.
|
| Some people can manage with slow network speeds at home, even
| though 100 Gbps single mode fiber is perfectly doable nowadays.
| And it's reasonable, because new SSDs do almost 120 Gbps.
|
| 1 Gbps made sense 20 years ago when single hard disks had
| similar performance. For some weird reason LAN speeds did not
| improve at the same rate as the disks did.
|
| But then again, I guess many could also still manage with 100
| Mbps connectivity at home. Still enough for 4k video, web
| browsing and most other "ordinary" use cases.
| noir_lord wrote:
| > For some weird reason LAN speeds did not improve at the
| same rate as the disks did.
|
| When it comes to wired, sending data 15cm is a very different
| problem than sending it 100M reliably - that and consumer
| demand for >1Gbps wasn't there which made the consumer
| equipment expensive because no mass market to drive it down,
| M.2 entirely removes the cable.
|
| I figured 10Gbps would be the standard by now (and was way
| off) and yet its not even the default on high end
| motherboards - 2.5Gbps is becoming a lot more common though.
| Avamander wrote:
| Most software and CDNs also don't utilise fast connections
| properly. It's kind-of a chicken and egg situation where
| hardware doesn't improve because customers don't demand it
| because software and services can't handle it (and you can
| start from the beginning).
|
| It is very slowly improving, but by far the fastest widely
| used services I've seen are a few gacha games and Steam
| both downloading their updates. Which is rather sad.
|
| Windows Update is slow, macOS update is abysmally slow,
| both iOS and Android stores also bottleneck somewhere. Most
| cloud storage services are just as bad. Most of these can't
| even utilise half a gigabit efficiently.
| walletdrainer wrote:
| > I figured 10Gbps would be the standard by now (and was
| way off) and yet its not even the default on high end
| motherboards - 2.5Gbps is becoming a lot more common
| though.
|
| All the new MacBook Pros come with 64Gbps wired networking.
|
| With an adapter you can also connect 100GbE, but that's not
| very special.
| Nextgrid wrote:
| 100Gbps over the LAN is unlikely to do you much good because
| not only is it expensive to get that kind of bandwidth end-
| to-end over the internet but most OS' network stacks and
| protocols (HTTPS/etc) are not efficient enough to take
| advantage of it (you will be bottlenecked by the CPU). So
| there is very little consumer and even business (outside of
| datacenters) demand for it because even just sticking a
| 100Gbps NIC and pipe in a consumer machine is unlikely to
| give you any more than 10Gbps anyway.
| threeducks wrote:
| A few things come to mind:
|
| - Games (400GB for Ark, 235GB for Call of Duty, 190GB for God
| of War)
|
| - LLMs (e.g. DeepSeek-V3.2-Exp at 690GB or Kimi-K2 at 1030GB
| unquantized)
|
| - Blockchains (Bitcoin blockchain approaching 700GB)
|
| - Deep learning datasets (1.1PB for Anna's Archive, 240TB for
| LAION-5B at low resolution)
|
| - Backups
|
| - Online video processing/storage
|
| - Piracy (Torrenting)
|
| Of course you can download those things on a slower connection,
| but I imagine that it would be a lot nicer if it went faster.
| hectormalot wrote:
| I have 1Gbit at home, but almost never reach those speeds
| when downloading games. It's one of those cases where it
| makes sense (I want to play now!), but I'm under the
| impression the limit is upstream (at steam most likely),
| rather than on my connection. (I do get those speeds on speed
| tests, doesn't seem to be my setup).
| Nextgrid wrote:
| ISPs happily collaborate with and put speed test servers in
| privileged locations on their network so you will get
| higher speeds there even if the actual peering to the
| outside world is much slower.
| hectormalot wrote:
| As I was typing this it came to mind. Will test against
| one of my own servers one of these days to confirm.
| Nextgrid wrote:
| You can try Fast.com (Netflix) or Cloudflare's one which
| are explicitly designed to work around this by serving
| the test data from the same endpoints the serve actual
| customer data, so ISPs can't cheat.
|
| This still doesn't guarantee however that you will
| achieve this speed to _any_ random host on the internet -
| their pipe to Cloudflare /Netflix may very well be fat
| and optimized but it doesn't guarantee their pipe to a
| random small hosting provider doesn't go over a 56k modem
| somewhere (I jest.. but only a bit).
| hdgvhicv wrote:
| Given that whether you get 30mbit or 30gbit from Netflix
| won't make a blind bit of difference it's not that useful
| a test. It doesn't do upload either as Netflix is all
| about consumption.
|
| Test to where you want to exchange high speed traffic.
| daveoc64 wrote:
| Fast.com does an upload speed test, but it's hidden
| behind the "Show more info" button.
| Hikikomori wrote:
| I get full speed on steam downloads, even set the limit
| lower so youtube doesn't buffer.
| Havoc wrote:
| Steam is tricky cause it has multiple potential
| bottlenecks. The steam cache, internet connection,
| decompression (i.e. cpu) and storage. Often hard to tell
| which limit you're hitting
| zamadatix wrote:
| I have 5 gigabit and usually get ~1.2 gbps, sometimes get
| up to ~2 gbps from Steam.
| toast0 wrote:
| You might check what region Steam is downloading from (it's
| in settings -> Download or something similar). If it's
| selected poorly, you might do better by picking one
| yourself.
| Namidairo wrote:
| > 400GB for Ark
|
| Ark is a strange case. It compresses very very well. Most of
| it ends up with compression ratios of around 80%.
|
| > Total size on disk is 628.32 GiB and total download size is
| 171.42 GiB.
|
| From SteamDB's summary of Ark's content depots.
| yatopifo wrote:
| I can't comment on the internet, but high-bandwidth wifi helps
| with VR streaming quality.
| amiga-workbench wrote:
| To not bottleneck the mechanical hard drives in their NAS, or
| to download games at a reasonable speed.
|
| Or even just work stuff, I've had to shift around several TB of
| 3D assets for my job while working from home.
| tonetegeatinst wrote:
| Homelab or they are into big data set usage.
|
| Or they seed large datasets for other researchers.
| ssl-3 wrote:
| It's nice to be able to do networked stuff with the network.
|
| 32GB isn't very big these days. In terms of cost, a decent
| cheeseburger costs more than a 32GB flash card does.
|
| A few months ago I needed a friend to send me a 32GB file. This
| took over 8 hours to accomplish with his 10Mbps upstream. 8
| hours! I felt like it was 1996 again and I was downloading
| Slackware disksets with a dialup modem.
|
| We needed to set up a resumable way to get his computer to send
| that file to my computer, and be semi-formal about it because 8
| hours presents a lot of time for stuff to break.
|
| But if we had gigabit speeds, instead? We could have moved that
| file in less than 5 minutes. That'd have been no big deal, with
| no need to be formal at all: If a 5-minute file transfer dies
| for some reason, then it's simple enough to just start it over
| again.
| tymscar wrote:
| All the replies you get here are totally valid, but Ill throw
| in another one.
|
| Why not? Life's too short anyways, and playing around with tech
| is one of those things that bring me joy.
| tills13 wrote:
| My anecdata is that I can download a steam or Xbox game
| (100+GB) in a few minutes. Plus it's just fun. High number =
| better and all that.
| NoiseBert69 wrote:
| Shuffle around RAW files if you are doing photography. These
| are 50-150MB files. A lot of them.
| ksec wrote:
| It is 2.5Gbps internet shared with everyone in the household.
| Once you start dividing by 4 or 5 people that number doesn't
| seems that impressive.
|
| It wasn't that long ago "internet" at home is literally just
| one person using it.
| Knork-and-Fife wrote:
| This is like asking why would anyone need more than a standard
| 110v North American electrical outlet in their home? Why would
| you ever install a higher capacity 220v socket somewhere?
|
| Because it's a utility and there's a wide world of use cases
| out there.
|
| For electrical maybe someone wants to charge an electric car
| fully overnight, or use a welder in their garage. Or use some
| big appliance in their kitchen.
|
| For Internet maybe they make videos, games or other types of
| data-heavy content and need to be able to upload and download
| it.
| qwertyuiop_ wrote:
| Running $60 Mikrotik HEX S 2025 and getting 1.2 Gbps on a "1G"
| connection !
| Nextgrid wrote:
| If that router has a 1Gbit port it's physically impossible and
| likely a measurement artifact.
|
| Actual speed on a 1Gbit port is something like 940Mbps
| according to experience (I believe the theoretical max there is
| 970).
| jonathanlydall wrote:
| Not sure what GP's situation is, but I have a 100Mb/s fibre
| internet package but all hooked up to 1Gbps capable equipment
| on my side.
|
| My typical speed test results are around 104Mb/s. Before
| being upgraded, on the 50Mb/s package I was getting 52Mb/s.
|
| My suspicion is that fibre network operator (OpenServe in
| South Africa) applies rate limits which are technically a
| little above what their customers are paying for, perhaps to
| avoid complaints from people who don't understand overheads.
| somehnguy wrote:
| 104mb/s is well under the theoretical max of 1gig
| networking, so you're truly just being limited by your ISP
| based on the plan you pay for.
|
| The poster above is claiming to see a physically impossible
| speed on 1gig networking.
| hdgvhicv wrote:
| The ISP sells a 100mbit package and delivers more than
| that, as the line speed will be higher and it's just
| policed in some fashion
| ssl-3 wrote:
| That's pretty typical. It's similar in the States:
| Spectrum, for example, generally overprovisions their
| connections a bit just because customer support is
| expensive to provide, and when things [ideally] work even
| better than advertised, support costs go down.
|
| And on that ISP side of things, it's a software-defined
| limit; it's just a field in a database or a config file
| that can be tuned to be whatever they want it to be.
|
| But the fellow up there says that they got 1.2Gbps through
| a Mikrotik Hex S: https://mikrotik.com/product/hex_s
|
| And that's just not possible*. The E60iUGS Mikrotik Hex S's
| own hardware Ethernet interfaces are 1000BASE-T, and it's
| simply not possible to squeeze more than 1.0Gbps through a
| 1000BASE-T interface. (It does also have an SFP port that
| it has one of is branded as "1.25Gbps," but reality is that
| it, too, is limited to no more than 1.0Gbps of data
| transfer.)
|
| *: Except... the 2025 version of the Hex S, E60iUGS, does
| have a 2.5Gbps SFP port that could be used as an ISP
| connection, and a much-improved internal fabric compared to
| the previous version. But the rest of its ports are just
| 1Gbps, which suggests a hard 1Gbps limit for any single
| connected LAN device.
|
| Except... Mikrotik's RouterOS allows hardware to be
| configured in many, many ways -- including using LACP to
| aggregate ports together. With the 2025 Hex S, an
| amalgamation could be created that would allow a single
| client computer to get >1Gbps from an ISP. It might even be
| possible to be similarly-clever with the previous version
| of the Hex S. But neither version will be able to do end-
| to-end >1Gbps without very deliberate and rather unusual
| effort.
| fulafel wrote:
| Wired Ethernet is typically full duplex.
| throwworhtthrow wrote:
| It's wasted effort in the US, since the 2025 budget bill directs
| the FCC to sell off much of the 6GHz band on which WiFi 7
| depends.
|
| https://arstechnica.com/tech-policy/2025/07/trump-and-congre...
| Nextgrid wrote:
| How would that work... will they have to force manufacturers to
| recall or issue mandatory updates to routers which already
| support it?
|
| FCC enforcement for interference can work for occasional
| troublemakers but there's no way they can go after every single
| consumer who (most likely not even realizing it) bought a 6Ghz-
| capable router that is encroaching on the now-privatized
| frequency band.
| rpcope1 wrote:
| I guess we're going to let AT&T, Verizon and everyone else just
| squat the entire spectrum. "5G" and the pillaging and theft of
| spectrum that seems to just sit idle anyways has been such a
| scam. If they wanted innovation there should be more ISM bands
| and less dependence being encouraged on wireless providers for
| "Internet access" as opposed to just biting the bullet and
| running more fiber and copper. But that would be bad for
| Verizon, T-Mobile and AT&T's bottom line so obviously we can't
| do that.
| Phelinofist wrote:
| Germany also wants to sell the 6Ghz frequencies to MNOs
| hhh wrote:
| 6Ghz for me causes random latency increases every 10 seconds or
| so, just had to stop using it because it drives me crazy
| martinald wrote:
| Mac?
| https://www.theregister.com/2025/10/23/apple_airdrop_awdl_la...
|
| It's been a problem for _years_. Basically the wifi card
| switches to another channel to see if anyone wants to do
| airdrop every so often. It's a bit of a joke to be honest that
| Apple still haven't fixed this.
| SG- wrote:
| the 2.5Gbit USB network adapters using Realtek driver are
| actually bugged on macOS and only max out at 1.9Gbit/sec or so.
| Sadly the solution has been to use non-realtek 2.5Gbit adapters
| or simply get the 5Gbit Realtek ones that sell for almost the
| same price.
| tymscar wrote:
| Are you sure? I just bought the Ugreen 2.5Gbps yesterday for
| this, and it uses a Realtek RTL8156BG chip. That's the one I
| used to get way above 2Gbps straight to the UDR.
| kalleboo wrote:
| I have a few cheap Realtek 2.5Gbps dongles spread around my
| house and get 2.3 Gbps TX and 1.9 Gbps RX (running iperf3
| --bidir to a LAN machine with 10 Gbps).
|
| Still beats Wi-Fi by a mile so I'm not complaining.
| SG- wrote:
| https://forums.macrumors.com/threads/anyone-seeing-speed-
| dro...
| Havoc wrote:
| Just checked mine...getting exactly gigabit speeds. Weird because
| everything should be on 2.5.
|
| Guess I need to do some debugging of my own
| kalleboo wrote:
| I get 1.6 Gbps line-of-sight no trouble with my U7 Pro, but I
| haven't managed to get MLO working. My only device with MLO
| support is the iPhone 16 Pro Max, and it refuses to connect to an
| SSID with MLO turned on...
| drnick1 wrote:
| > I recently upgraded from a UniFi Dream Machine to a UniFi Dream
| Router 7
|
| What do these devices do that can't be accomplished by an OpenWrt
| One + an external AP for less money and fully FOSS?
|
| Another option would be a mini-PC running Linux, but it's perhaps
| overkill for a domestic router.
|
| Edit: Actually the OpenWrt One does have built-in WiFi, so you
| don't even need the external AP.
| ishanjain28 wrote:
| Good band steering and roaming. I would not use openwrt if I
| had to use multiple APs to cover the area.
| rubenbe wrote:
| OpenWRT does support 802.11r fast roaming for multiple APs.
| The problem with OpenWRT is/was the configuration of multiple
| APs. There is OpenWISP, but they mostly target very large
| setups (>100 APs). So I built OpenSOHO using the OpenWISP
| daemons on the AP and a pocketbase frontend.
| (https://github.com/rubenbe/opensoho). No band steering yet
| unfortunately.
| riobard wrote:
| > What do these devices do that can't be accomplished by an
| OpenWrt One + an external AP for less money and fully FOSS?
|
| Nice UI (as the company is best known for https://ui.com)
| NoiseBert69 wrote:
| (TLDR: if you want to use bleeding edge technology you must use
| bleeding edge drivers and firmware blobs)
|
| We have tested WiFi-7 gear in our lab: from the cheapest TP Omada
| EAP783 to the latest most expensive Cisco AP+Controller.
|
| Our findings:
|
| - Driver quality from the modems is still below average on Linux.
| If you want to test Wifi-7 go with the Intel BE200 card - most
| stuff works there. Warning: this card does not work with AMD
| CPUs.
|
| - We have seen quite a bit of problems from Qualcomm and Mediatek
| cards. Either latency issues, weirdo bugs on 6GHz (not showing
| all SSIDs) or throughput problems
|
| - Always go with the latest kernel with the freshest firmware
| blobs
|
| - MLO is difficult to get running properly. Very buggy from all
| sides. Also needs the latest version of wpa_supplicant -
| otherwise it will not come up. And be aware: there are several
| MLO modes and not all of them offer "two links for twice the
| bandwidth".
|
| Also expect to hit problems from AP side. If you read the TP
| Omada firmware changelogs you see that they are still struggling
| with a lot of basic functionality. So keep them updated to the
| latest beta versions too.
|
| I use a Qualcomm QCNCM865 in my privat setup with an AMD CPU.
| Feels like the latest firmware blobs and kernel drivers brought
| stability into their components.
| NooneAtAll3 wrote:
| > Warning: this card does not work with AMD CPUs.
|
| what causes that? (I have no idea how wifi cards work)
| bcook wrote:
| I'm just guessing, but I would say Intel CNVi.
| https://en.wikipedia.org/wiki/CNVi
| NoiseBert69 wrote:
| That's the reason.
|
| But I can confirm that the Intel BE200 works with the
| popular Intel n100/n305 Mini Computers.
| ksec wrote:
| As I have been trying to tell the world and keep repeating. The
| best version and implementation of WiFI 6E is WiFi 7. So if
| anyone want decent WiFi 7 they will have to wait till WiFi 8.
| rcarmo wrote:
| Went through a similar tuning process with Wi-Fi 6 on OpenWRT
| recently: https://taoofmac.com/space/reviews/2025/09/14/1630
|
| In my case, I forgot I had to change encryption type to associate
| at higher speeds.
| tymscar wrote:
| Oh, good point. That was actually the first thing I missed, but
| when I created a new Wi-Fi 7-only SSID, Unifi wouldn't let me
| pick anything lower than WPA3 if I only used 6 GHz. So that
| sort of fixed itself.
| mstaoru wrote:
| I brought my WiFi 7-capable ASUS RT-BE96U to Germany (from China)
| and I proudly notice that my average download speed is up to ~105
| Mbit from ~95 Mbit with the stock Vodafone router.
|
| "Silicon Valley of Europe", my a*s.
| fulafel wrote:
| The wifi bottleneck is such a tragedy. We'd otherwise have 10G
| home broadband probably.
| mannyv wrote:
| There are times when low power is better, because it allows the
| router to ignore far away clients.
|
| In simple terms, far away = more work to communicate = more
| airtime = less throughput.
|
| It probably only matters with multiple devices.
___________________________________________________________________
(page generated 2025-11-02 23:01 UTC)