Interpretation of speed test results

[TheLyppardMan]

I wonder if anyone might be able to interpret some speed test results I have obtained on my new BT FTTP 500 Service?

Using my RT-AX88U with QoS turned off (just be be sure of no additional interference), I'm getting about 385 down and 73 up via Wi-Fi. I've used the router's own speed test option in the GUI and also check it against BT's own speed tester (www.speedtest.btwholesale.com). Both methods return very similar results. However, if I connect the supplied BT Smart Hub 2 and connect my laptop to the hub via an Ethernet connection, I get around 498 down and 55 up from the BT test site. My query is why is the download so much faster on the BT web site and yet the upload is quite a bit slower when using the hub? I could understand if both were slower or faster, but these results seem a bit odd at first glance. Any ideas what might be going on here? I can try some additional experiments if necessary, but it'll have to wait until the new year as my children are at home all day now and they can't of course manage without the Internet for more than a few minutes LOL.

 

[TheLyppardMan]

Here's the router's speed test results for the last 24 hours.

 

[AndreiV]

Simple , you are using ethernet for one test and wifi for the other.

You need to test with one connection type against the same test server with both routers to get a comparison.

 

[TheLyppardMan]

Simple , you are using ethernet for one test and wifi for the other.

You need to test with one connection type against the same test server with both routers to get a comparison.

Well I don't think it's quite that simple. Why would I get lower upload speeds via Ethernet than via Wi-Fi for the upload test but higher speeds on the download test? I would understand that if both speeds were lower on the Wi-Fi for instance, but that wasn't the case. I got about 498 Mbps download via a wired connection to the BT hub but only around 55 Mbps upload and yet on Wi-Fi via my ASUS RT-AX88U I managed to get about 73 Mbps upload but only about 425 Mbps or so on download (with QoS turned off).

 

[adrenalize]

What device are you using on WiFi to do the tests? What WiFi link speed have you got? Whilst the AX88U might be capable of faster speeds often the clients can't.

I have an AX88U and due to have 500/75 installed a week into Jan so will see how mine compares. I did see another thread where someone changed their MTU and MRU which improved throughput although they were on 900.

 

[eightiescalling]

What do you get if you do a test on the AX88U over ethernet? There will be differences in the AX88U vs BT Hub on how they work so it would still need to be an ethernet vs ethernet test for comparison. To be honest, while Wifi can theoretically cope with those high speeds I'd do any speed testing on ethernet just for the certainty over the connection.

I tend to use the ThinkBroadband speed test - that also gives a couple of other indicators of connection health which can sometimes contribute to issues (not to the degree you're seeing though).

 

[L&LD]

Yes, it is that simple. Your tests must be consistent and make changes to only one part of the setup, to compare, or else ('noisy' data).

Many reasons why the up/down figures don't track together. Minimize those variables and the data will make more sense.

 

[TheLyppardMan]

Yes, it is that simple. Your tests must be consistent and make changes to only one part of the setup, to compare, or else ('noisy' data).

Many reasons why the up/down figures don't track together. Minimize those variables and the data will make more sense.

So if I've understood correctly, it's perfectly possible to get much higher upload speeds over Wi-Fi using the ASUS router than using an Ethernet connection via the BT hub but the reverse is also true for the download speeds. It still seems a bit weird to me, but I'll try some like for like tests after the Christmas break and post the results into this thread.

 

[dosborne]

It isn't just a matter of the variation in wired vs wireless, it is also wired link speed (since you didn't specify), wireless link speed, processor, driver optimization, and a large number of other things when you change devices and try and compare results

 

[AndreiV]

You also need to consider what type of speed test the different providers are using. Many sites use single thread , others may be multi thread, the results will be different.

You can only compare like for like. You have membership at Thinkbroadband , look at their speed tests , they have single and multithread tests, the best tests available there are file downloads of specific sizes which you can time.

 

[TheLyppardMan]

OK, so how about this for a plan. Run an Ethernet cable from my laptop to the BT hub and conduct 2 or 3 speed tests, one set of tests using the BT website and another set using a different speed test website, taking the average of each set of results . Then disconnect the cable and repeat the tests over Wi-Fi. After that, repeat both sets of tests using my ASUS router, perhaps adding an additional set of Wi-Fi tests (with ax switched off, as the BT hub doesn't have a Wi-Fi 6 option)?

 

[Kingp1n]

OK, so how about this for a plan. Run an Ethernet cable from my laptop to the BT hub and conduct 2 or 3 speed tests, one set of tests using the BT website and another set using a different speed test website, taking the average of each set of results . Then disconnect the cable and repeat the tests over Wi-Fi. After that, repeat both sets of tests using my ASUS router, perhaps adding an additional set of Wi-Fi tests (with ax switched off, as the BT hub doesn't have a Wi-Fi 6 option)?

How old is your ethernet wired device? Using an older PC/laptop(using old drivers) as others have mentioned will provide substandard results.

I had the issue awhile back using a new samsung cell phone (wifi speed testing) and comparing them with an old laptop that was wired. Results were Great on wifi on new devices not so much on the laptop.

 

[TheLyppardMan]

How old is your ethernet wired device? Using an older PC/laptop(using old drivers) as others have mentioned will provide substandard results.

I had the issue awhile back using a new samsung cell phone (wifi speed testing) and comparing them with an old laptop that was wired. Results were Great on wifi on new devices not so much on the laptop.

My laptop is about 18 months old. These are the specs:-
2.60 gigahertz Intel Core i7-10750H
384 kilobyte primary memory cache
1536 kilobyte secondary memory cache
12288 kilobyte tertiary memory cache
64-bit ready
Multi-core (6 total)
Hyper-threaded (12 total)

 

[L&LD]

Network Performance Testing Best Practices.

A wired/wireless testing technique (overview).

• Keep good notes.
• Use more than one device.
• Test more than one location.
  • (True for both wired and wireless test runs, particularly if wired speeds fluctuate between the different devices used).
• Test using more than one test site.
  • (But be sure to use the same servers between test sites, if possible. If not possible, merely note the server used).
• Do not average the results above.
  • Averages are useless here, as they are not valid data points.
  • Comparing like for like is what you're interested in, over time (and/or when changing or tweaking settings).

Here is what I do at some customers' locations. Where possible, I test wired connections also (and not just WiFi (both radios/bands, and in the near future, three radios/bands or more).

1. Change (temporarily) the main SSIDs and password (and I use new/unique SSIDs each time).
2. Unplug all LAN-connected equipment (at their end, not at the router location).
   1. You want no other devices using bandwidth and skewing the test results you'll get below.
3. I do not change/tweak any WiFi settings until I get a 'baseline' with the above and the following steps, done.
   1. i.e., this is their network at its 'best', when I first see/encounter it.
4. I normally perform the following tests with my phone (Note 10 Plus), my Intel AC-equipped, 'low-powered' laptop, and my Intel AX210 equipped workstation laptop.
   1. My devices are kept fully updated and are a known quantity (this makes it easy to quickly test for customer device-related issues too).
5. Except for the phone, the testing with the laptops is performed in 'Best Performance' mode, to eliminate power-saving performance variability (or ideally, when the laptops are plugged into AC power and also on 'Best Performance' mode too).
6. At least two 'main' usage areas are tested with each device above. The specific locations are noted and used consistently between tests, including whether it was used sitting/standing, the direction faced, and the orientation of the device/screen and the user, relative to the router and device being tested.
7. I use fast.com, speedtest.net, and the speed test site of the customer's ISP.
   1. I also just browse the 'net with two or three (personal) websites to get a feel of the network, overall.
8. I record the information in a spreadsheet with the following columns.
   1. Customer, date, time, Router Model, Config (1, 2, 3, etc.), Down (Mbps), Up (Mbps), latency (ms, unloaded, loaded), test location, and website 1, 2, 3 (responsiveness).
   2. Config 1 or 2 or 3, etc. will be detailed further in my notes (same spreadsheet). Including firmware version and what settings past defaults (or existing) were changed.
9. With all information recorded as indicated above at all locations and with all devices.
10. Do not use WiFi testing apps when doing the above. Do not even have them running (they will affect results). Do keep great notes!

The proper analysis of the data collected above is to compare the same device used at the same location to the same testing site using the same server (for that site). Anything else is just tossing a coin.

When using the above method to dial in your WiFi (and other) settings, the maximum Down/Up speeds are not the prime determining data points to use/choose. Rather, the last columns are ('personal' website 1, 2, 3, responsiveness - step 8.1 above).

I hope the above is useful. Please ask for any clarification of the quick overview of the beginning of my network-tuning process, above.

 

[TheLyppardMan]

Network Performance Testing Best Practices.

A wired/wireless testing technique (overview).

• Keep good notes.
• Use more than one device.
• Test more than one location.
  • (True for both wired and wireless test runs, particularly if wired speeds fluctuate between the different devices used).
• Test using more than one test site.
  • (But be sure to use the same servers between test sites, if possible. If not possible, merely note the server used).
• Do not average the results above.
  • Averages are useless here, as they are not valid data points.
  • Comparing like for like is what you're interested in, over time (and/or when changing or tweaking settings).

Here is what I do at some customers' locations. Where possible, I test wired connections also (and not just WiFi (both radios/bands, and in the near future, three radios/bands or more).

1. Change (temporarily) the main SSIDs and password (and I use new/unique SSIDs each time).
2. Unplug all LAN-connected equipment (at their end, not at the router location).
   1. You want no other devices using bandwidth and skewing the test results you'll get below.
3. I do not change/tweak any WiFi settings until I get a 'baseline' with the above and the following steps, done.
   1. i.e., this is their network at its 'best', when I first see/encounter it.
4. I normally perform the following tests with my phone (Note 10 Plus), my Intel AC-equipped, 'low-powered' laptop, and my Intel AX210 equipped workstation laptop.
   1. My devices are kept fully updated and are a known quantity (this makes it easy to quickly test for customer device-related issues too).
5. Except for the phone, the testing with the laptops is performed in 'Best Performance' mode, to eliminate power-saving performance variability (or ideally, when the laptops are plugged into AC power and also on 'Best Performance' mode too).
6. At least two 'main' usage areas are tested with each device above. The specific locations are noted and used consistently between tests, including whether it was used sitting/standing, the direction faced, and the orientation of the device/screen and the user, relative to the router and device being tested.
7. I use fast.com, speedtest.net, and the speed test site of the customer's ISP.
   1. I also just browse the 'net with two or three (personal) websites to get a feel of the network, overall.
8. I record the information in a spreadsheet with the following columns.
   1. Customer, date, time, Router Model, Config (1, 2, 3, etc.), Down (Mbps), Up (Mbps), latency (ms, unloaded, loaded), test location, and website 1, 2, 3 (responsiveness).
   2. Config 1 or 2 or 3, etc. will be detailed further in my notes (same spreadsheet). Including firmware version and what settings past defaults (or existing) were changed.
9. With all information recorded as indicated above at all locations and with all devices.
10. Do not use WiFi testing apps when doing the above. Do not even have them running (they will affect results). Do keep great notes!

The proper analysis of the data collected above is to compare the same device used at the same location to the same testing site using the same server (for that site). Anything else is just tossing a coin.

When using the above method to dial in your WiFi (and other) settings, the maximum Down/Up speeds are not the prime determining data points to use/choose. Rather, the last columns are ('personal' website 1, 2, 3, responsiveness - step 8.1 above).

I hope the above is useful. Please ask for any clarification of the quick overview of the beginning of my network-tuning process, above.

Thanks for that, but I don't envisage going into such detail, as I'm only aiming for an estimate of how my laptop compares when using the RT-AX88U v. the BT Hub and I don't expect to be using the BT Hub as a matter of routine, as it doesn't have anywhere near as many features as my ASUS router, nor would I expect it to. I'm just a bit curious really as to how they compare in terms of everyday performance from an end users' point of view. I'll see how it goes and update this thread after the festive break is over.

 

[Tech9]

I'm just a bit curious really as to how they compare in terms of everyday performance from an end users' point of view.

Both will provide very similar user experience in common Internet use. They may have different range and maximum transfer speeds, but usual web browsing, media streaming, work/learn from home, etc. fits under 100Mbps speeds. Asus advantage is in more user accessible features. Most users will never notice any difference and would not care much about extra features and more often firmware updates.

 

[L&LD]

You may not need to go into that kind of detail, but the overall points are still pertinent.

Bottom line is compare like to like. Keeping good notes helps achieve that.