Odd Speed Test Results

Hi,

Just got BT Fibre 500 installed today and have been doing some speed tests on my PCs.

On my ASRock J5040-ITX based system I get the expected result



But on my ASRock J3355B-ITX based system and other PCs I only get a lower result like this



Is this because high speed Broadband speed tests are very much dependant on CPU power ?

Both systems were tested with CAT5E cable connected directly to the BT SmartHub2
and both systems have the same Realtek PCIe GbE Familt Controller NIC and use the same driver V10.35.510.2019 driver running Windows 10 Pro. I checked the NIC settings on both systems and as far as I can tell they are identical.

Another system - an Acer E15 laptop running a Celeron 2940 CPU but also woth a Realtek Gb NIC gives the same result as the J3355B-ITX system

Only my newest ASRock J5040-ITX system gives me the expected result.

Any ideas what is going on here?

Regards,
Gus

Hi Gus,

In reply to a post by gussie5555:

On my ASRock J5040-ITX based system I get the expected result


But on my ASRock J3355B-ITX based system and other PCs I only get a lower result like this


Is this because high speed Broadband speed tests are very much dependant on CPU power ?

Both systems were tested with CAT5E cable connected directly to the BT SmartHub2
and both systems have the same Realtek PCIe GbE Familt Controller NIC and use the same driver V10.35.510.2019 driver running Windows 10 Pro. I checked the NIC settings on both systems and as far as I can tell they are identical.

Any ideas what is going on here?

Speed tests will take up some CPU power but it's far more likely the issue with the setup of the browser, anti-virus software (which might hijack web traffic and cause buffering or similar), or even network card MTU - Our test is very sensitive, i.e. it it drops it shows it quite quickly whilst some others will be more forgiving (i.e. they look for a more 'peak' speed than average.. We do try to strike a balance not to ever over-read (it's not impossible because of how streams can be interfered with by sofware).

Your specific profiles look very different. The 500Mvps results are very static whilst your 200Mbps results vary a lot. Both are using IPv4 but also your latency jumps from 12 to 22ms. This is why I would double check the wi-fi use and rule that out..

Are you sure you've got Wi-fi disabled on the second device? I sometimes find on Mac, wi-fi takes preference over physical when you don't expect when it's connected to the same LAN.

Hi Seb,

Thanks for replying.

I can rule out any potential Wi-Fi interference since my ASRock J5040-ITX and J3355B-ITX have no Wi-Fi interfaces present - only Gb lan interfaces.

It's a real mystery that only one of my PCs gives the expected result not just from ThinkBrodaband but other speed test sites like Fast,com. Note I have repeated these speed tests several times and get consistent results.

Dug out my old Gigaabyte GA-X58A-UD3R PC running an Intel i7-930 cpu (again no Wi-Fi only Realtek RTL8111E Gb lan) and got this result



Just to repeat only my ASRock J5040-ITX gives me the expected Fibre 500 results.
And yes I have tried using Chrome instead of Firefox - same results.

A real mystery!

Gus

* Try booting into an Ubuntu live USB image. Linux often works far better than Windows at networking.
* Instead of a browser-based test, try using a speedtest app, or a command-line speed test tool. Ookla (speedtest.net) provide both of these.

For example, on my Macbook Pro 2015, using speedtest.net in a browser saturates the CPU at 100% (even on a 300M line), whereas the speedtest app is much more efficient and leaves plenty of CPU unused.

Do you have anti-virus software? Does it differ between the two?
This is one of the sources of problems we see because they interfere (by design) with web traffic at times.

No AV software other than the MS stuff - standard settings on all of my PCs.

I'll try your sugestions - but the question remains why one of my PCs consistently reports expected results whilst the others are reporting much lower speeds - and this is using browser based tests.

I get exactly the same resukts when I replace the BT SmartHub2 router with my ASUS RT-AC68U router.

In reply to a post by gussie5555:

No AV software other than the MS stuff - standard settings on all of my PCs.

It might be CPU I guess then .. Would be useful to see what you can see with other non web tests.

Thank You candlerb you have come up trumps!

I downloaded the Ockla Speetest app for Windows and ran it on my Acer E3-112 Intel Celeron CPU N2840 (2.16GHz) laptop using Cat5E cable to my Asus RT-AC68U router and got this tonight

[img]https://i.ibb.co/4VP455s/260423-1930-Acer-E3-112-Int...[/img]

If I use the ThinkBroadband test I get this



And this is with my slowest machine - so the answer is to use the Ockla Speetest app wgere possible. I guess browser based tests are very much dependant on CPU power - maybe my J5040 system just has that bit extra oomph to cope with the browser loading.

Anyway thanks again to all for clarifying the issue. Once you reach very high speeds I guess you have to take care in interpreting speed tests.

"Once you reach very high speeds I guess you have to take care in interpreting speed tests. "

Or not worry about them if your internet use is not affected.