Broadband News

Why its important to use an Ethernet connection when speed testing

The advent of the smart phone and tablet meant the days of having to sit in front of a massive beige box to access the Internet are a memory for millions but for broadband providers the variability of Wi-Fi performance means the scope for complaints about slow speeds are even higher today.

Our broadband speed test will run on computers, tablets, phones, games consoles and some smart TVs and while we recommend people connect with an Ethernet cable where possible many people don't and by looking at Virgin Media speed test results we can get a good idea the impact on your experience when using different devices.

The 10 second summary is this, even though many people use a laptop or PC with a wireless card PC based speed tests are consistently faster than mobile phones which in turn are consistently faster than a tablet.

Plot of Virgin Media download speeds on computers, mobiles and tablets
Trend of average download speeds on different devices for Virgin Media
Plot of Virgin Media upload speeds on computers, mobiles and tablets
Trend of average upload speeds on different devices for Virgin Media

The download chart is showing that in November 2017 the mean download for PC devices was 68.5 Mbps, phones 55.3 Mbps and tablets 44.1 Mbps and for upload speeds the figures are 8.3 Mbps PC, 6.4 Mbps phone and 5.5 Mbps tablet. We have not included the charts but the quality metric which is looking at how stable a test speed is for every test indicates PC devices score 0.6 (Grade A) versus 1.0 (Grade B) for mobile phones and 1.2 (Grade C) for tablets, or put another way as the quality score increases the chances of you enjoying the buffering symbol increase.

Why is this important? Beyond bragging rights about the fastest connection in your street. Well with the increasing use of smart phones and tablets as default web access and video streaming devices it is likely that a good chunk of the complaints around broadband issues are related to Wi-Fi rather than the underlying broadband access method, be that a piece of string, a copper pair, coax or a fibre optic strand. In the Virgin Media November speed stats we are using in this article 57% were found to be on a PC, 30% using a mobile phone and 13% with a tablet.

We suspect that the reason tablets may be slower is that people tend to upgrade their mobile phone in line with their mobile contract term, whereas tablets you keep for longer so there will be more devices with less advanced Wi-Fi features.

TV adverts like to hang Hollywood stars from helicopters or show technicians unplugging a cable to showcase their Wi-Fi routers but the reality is that if you want to have the best experience when streaming films to your TV if at all possible connect via an Ethernet cable and if that is not possible make sure the device is using the 5 GHz band on your wireless router.

A top tech tip is that while broadband providers and router manufacturers configure both the 2.4 GHz and 5 GHz wireless bands with the same network name (SSID) to actually get the best performance change the names so you can identify them and just teach your devices the credentials to access the 5 GHz band (this is assuming the 5 GHz band will reach where you want, i.e. 2.4 GHz has a longer reach).

Comments

It is an interesting issue as I suspect there are an increasing number of households that will have no devices that have an Ethernet port. Mobiles and tablets tend not to have them. Thin laptops (Apple but also other makes such as Dell) have no Ethernet port because the port is too large to fit in the chassis. So, the only way people could test would be buying something extra that they would probably only use for speed tests. Going forwards it will become less realistic to expect a household to have a wired device to test from.

  • ian72
  • 14 days ago

https://www.bobjgear.shop/adapters?lightbox=dataItem-irmph699

  • adslmax
  • 14 days ago

@adslmax Yes of course there are adapters available. Most households won't have one though and if asked by an ISP to run a wired test before they can log a call they would have to source an adapter somewhere, and they would have to work out what they need first (most people wouldn't have a clue and will an ISP advise them?). Looking forwards more people will have connections over 100Mb - that means their wireless may actually be faster than their wired network - would need a gigabit USB adapter. I agree to use wired but it isn't necessarily a simple option in a growing number of homes.

  • ian72
  • 14 days ago

There's something just not right with thinkbroadband's speed test.

On wifi I'll get around 160Mb/s or so results here but other site's like DSLreports and speedtest.net I'll get upwards of 300Mb/s on my virgin media Vivid 350 package while downloading apps from Google Play I'll get results of 400Mb/s! Over wifi.

What's up with that?

  • Purplemonkeyspa
  • 14 days ago

Nothing wrong with the test, but Virgin Media seem to send the traffic over a congested path within their network.

Additionally our tester is designed to be more diagnostic e.g. displays median speed rather than a higher percentile which some do. also some testers use massive numbers of downloads to saturate a connection. We do a single test and a multiple test, these all lead towards the test sometimes highlighting local congestion too.

  • andrew
  • thinkbroadband staff
  • 14 days ago

http://tbb.st/1512128267589763155 the shape of the single test (tbb) shows its taking that connection longer to saturate.

http://tbb.st/1512130208849133955 this user has it worse, but does get decent numbers from the test

http://tbb.st/1512102068866024255 a 200 Mbps user nailing it.

Got a link to your results?

  • andrew
  • thinkbroadband staff
  • 14 days ago

Here are the few tests:

https://www.thinkbroadband.com/speedtest/1511896157805469955

https://www.thinkbroadband.com/speedtest/1511896118367593355

https://www.thinkbroadband.com/speedtest/1511896068740637655

All wifi.

  • Purplemonkeyspa
  • 14 days ago

@Purplemonkeyspa

You're not alone! On my Openreach based FTTP connection (330/30), I pretty much get the full whack 310 Mbps on Speedtest.net, dslreports, testmynet etc. Yet bizarrely this is what TBB gives me (over wifi but wired is only a little better)

https://www.thinkbroadband.com/speedtest/1512152959102242655

  • baby_frogmella
  • 14 days ago

Tester is happy to go faster with FluidOne e.g.

http://tbb.st/1504105283781367855
http://tbb.st/1497960105642743955 near perfect result for FTTP
http://tbb.st/1502970339460435555

  • andrew
  • thinkbroadband staff
  • 14 days ago

For Purplemonkeyspa something is not right for you, the same shape graph throughout when what you really want is http://tbb.st/1512130664669971355

A very sharp rise up and to hold the speed.

More nice ones http://tbb.st/1512151085308581955
http://tbb.st/1512147135495238455
http://tbb.st/1512144046563840855
http://tbb.st/1512147068483292755

So its not the tester itself, and as all Virgin customers come to us the same way if was our network you'd all have the same slow ramp up pattern.

It may be multiple routes or regional issues in Virgin network giving varying results

  • andrew
  • thinkbroadband staff
  • 14 days ago

Why does this not affect speedtest.net?
Because they have test files on the virgin network and the picking algorithm will pick lowest latency thus working around congestion a lot of the time.

Why does dslreports/testmy.net not show it? Because they grab up to 32 files at once from multiple locations, so once again if one or two routes out of Virgin are slow it does not matter.

  • andrew
  • thinkbroadband staff
  • 14 days ago

I agree with others. TBB speed test isn't reliable one. Worst ever. I don't use it anymore.

  • adslmax
  • 14 days ago

So when others can get

http://tbb.st/1512094535629323555
http://tbb.st/1512118425300481555
http://tbb.st/1512135619382535455
http://tbb.st/1510840704201318155
http://tbb.st/1511437233535901855
http://tbb.st/1510614327248942755
http://tbb.st/1509619529582699155

It is classed as the worst ever, perhaps I should resign and go and live in a tent in the park

  • andrew
  • thinkbroadband staff
  • 14 days ago

@andrew

Thanks. I honestly had no idea of the technicalities. I've never had it explained before so I just assumed your TBB's test was a bit wonky.

That said I KNOW my phone is capable of my line's max speeds because I've seen it. 400-480Mb/s downloading apps from the Play Store. Results I know are correct because I timed them and used an online DL time calculator to confirm the "50M-60MB/s" shown in the progress bar. So, wifi isn't TOO bad with the right conditions.

Unfortunately I don't own an ethernet capable device apart from my PS4 Pro which has it's own problems with Sonys DL servers

  • Purplemonkeyspa
  • 14 days ago

https://www.thinkbroadband.com/broadband/monitoring/quality/share/37b51ff102c9887919b30f26f5eb8e9ee4a13875-01-12-2017

That's my BQM. As you can can see it's pretty gross. Could that be affecting anything?

  • Purplemonkeyspa
  • 14 days ago

@MrSaffron (NO BT FTTP or G.Fast speedtest result?)

  • adslmax
  • 14 days ago

  • MCM999
  • 14 days ago

Gfast
http://tbb.st/1512068770190359555
http://tbb.st/1511806041170860755
http://tbb.st/1510255153770465455

FTTP
http://tbb.st/1510473510728775455
http://tbb.st/1510040968112642155
http://tbb.st/1511855539200186855

  • andrew
  • thinkbroadband staff
  • 13 days ago

Virgin Media ping performance is impacted by Intel Puma chipset issue in the SuperHub 3.0 (not just BQM but actual ping for gaming too) so using a SuperHub 2.0 is current solution

  • andrew
  • thinkbroadband staff
  • 13 days ago

I look forward to the day WiFi will have any effect on my speed, as a demo to a friend who was worried about moving to a house with a inconvenient master socket (+ in a more rural area) I tested with Ethernet cable and by WiFi as far from router as possible whilst still in the house, there was no difference.

  • burble
  • 13 days ago

My wired is the same as wireless, on 5Ghz with wireless AC. I can get 300Mbps+ over wireless, and on my WiFi boosters (linked via 5Ghz) 170Mbps+.

  • ukhardy071
  • 13 days ago

Fully aware that some people do get better speeds, and the paragraph at the end about splitting bands will help a lot of people get better BUT the numbers point towards the overall average being impacted.

So all those doubting what I've said are showcasing what will be a problem once the new broadband speeds advertising regime starts, i.e. doubt about what a provider is saying and may lead to less people switching.

  • andrew
  • thinkbroadband staff
  • 12 days ago

So instead of bleating about my speed this and my speed that, what is a real world solution to speedtest when households are all wi-fi...

One could be to have the speed test code as part of the firmware build in the router with the option to use providers test or allow 3rd party test. This could then be accessed via a web browser (begone flash..)

  • mavrik64
  • 11 days ago

Router embedded solutions is one solution, but then you get the scenario where the public don't understand why their expensive tablet is not streaming well when the speed test is showing a good result.

We have also had people very confused when they use one of those solutions, as they think it is running on their tablet, since the web page is viewed on the tablet.

Like so many tech areas there is a lot of education to be done still, e.g. the difference between B and b still stumps some and that includes the copy writers for some providers at times.

  • andrew
  • thinkbroadband staff
  • 11 days ago

Post a comment

Login Register