Speedtest not showing correct speeds

Hi , I’m using the Speedtest.net integration which apart from stopping from time to time doesn’t display the correct speeds for download or upload. I have Virgin Media with 264mbps down and 35mbps up and Speedtest shows 23 & 17.

Is it limited to a certain speed or is there another reason.

Also tried the FastCom integration which shows 30mbps … so different but still not right

Any ideas please … thank you

Just to add I’m using HA on an Homix box which I believe as Gbps nic and my home network for wired is Gbps.

It’ll be limited to the maximum speed of the hardware. What is a “Homix box”?

Do you need to test your speed? Have you been having problems with it?

Hi, the HomeIX box is provided by Vesternet with built in Zigbee and Zwave and HA . So far a brilliant little box all be it expensive

I believe this box as a Gbps i/f as I said in the post so I cant see that being a limiting factor.
Not having problems with Internet but that’s why I monitor to ensure I am aware of problems when they occur.

Regards

Values don’t quite add up, but I’ll say it anyway - it’s easy to confuse KB/sec / MB/sec with Kb/sec / Mb/sec. (bytes and bits)

Have you tested from another device plugged straight into your router with a cable?

Having a gigabit NIC isn’t the same as being able to handle the traffic at that speed. Box capability would be my prime suspect.

I have to concur, the speed test is pretty far from off. My spectrum circuit is a 1GB/s download and 40MB/s upload.

On my laptop, wired into the wall, I get 905.3MB/s down and 32.74MBs up.

On my Raspberry PI 4b running HA, I my last reading was 132.64MB/s down and 31.41MB/s up.

There are many limitations and factors to be taken into account.
First is hardware, where the bandwidth inside the devices also plays a part, like when on many raspis that are actually limited to around 300Mb/s internally, so a 1Gb/s NIC will never be fully usable.
Then there are the limitations on the networks, both internally and externally, which can be different MTU sizes, overburdened devices, like many or complex firewalls rules on a small SoHo router.
And finally hardware and networks will be shared, so when other stuff is using the those, then your results will be skewed and maybe even skewed so much that they are useless.

1 Like

I do agree the hardware can be a factor but when running several speed tests from my phone I get around 490mbps with (depending on time of day ) anything from 87mbps to 260 upwards so even that is higher that what Speedtest is reporting

I know the attached is quite old but I have been running it for a long time now. I followed this as the built in speedtest to HA was not giving the right speeds for me (and others in the attached thread).

With this installed I get the speeds I expect

I also saw an openspeedtest addon a while back, not tested that though

Just done a speedtest at my router and got 871Mbps/178Mbps. My service allows 900Mbps/180Mbps. The method above in HA gets me 869Mbps/170Mbps, pretty close

I would recommend using the official SpeedTest CLI Binary.
Here is a Community Thread with a couple of options for how incorporate this into HA

Don’t forget that you are throttled by your lowest connected ethernet. Your wifi might report well if the wifi is either on or directly connected to the router via a Cat6e cable. But, for instance, if you have a GB card on the computer you are testing that goes into a hub that is only 10/100 or that hub is using a lower speed cable to connect to the router (or basically any slower hop along the way) then you are going to test at the lowest speed in the path.

I’ve had a lot of folks ask me why they just upgraded their computer to GB ethernet to see no difference in speed, not realizing that in the wall was Cat5 cable, which has a 100Mbps speed, thus rendering their new GB card totally useless.

That, on top of the hardware, could be the difference. Also Speedtest sends fairly massive payloads to test speeds, I learned this the hard way on my cellular Internet when I was doing speed tests regularly, I burned through my plan in a day - that’s to say that you need a decent processor and drive speed to be able to really report accurately.

:frowning:
greed but my lan is hardwired from HA box via gig switch to gig port on router , so only possible bottlenecks are HA box , switch or router . All new hardware so can’t see that being the problem.

Also if there was a hardware limit I would expect to see a much higher result even if it’s not the correct one . Even with my old router which was still gig port I wasn’t getting correct speed readings . Will think of something else …:frowning:

Ya’ think??
Is it too late to return it?

Have you opened it up to see what hardware is inside?

On the topic, there’s way too many variables to determine where the bottleneck is. Have you connected a PC by Ethernet directly to the router to get a baseline speedtest?. Run Speedtest multiple times and average the results. Also use a different speedtest host a few times. Once you have the baseline move the PC to the switch going to your Homeix hub. Test again.

New is not equal with powerful.
There are being made plenty of new underpowered routers today.

Hook up your laptop directly to the cable modem and run a speed test on it. That’ll tell you what the real connection throughput is at that moment. Note that cable speeds are “up to” and can be much lower depending on area load.

Try a speedtest from your computer hardwired to the switch. Note the results, and if they are less then what you expect do the same from the router. If you get expected speeds both places, the issue is with your Homeix. If the speeds are lower then expected on your switch, try switching ports on the cable going to the router, or replace that cable.

You all have correct ways to diagnose this, but it is a bit unorganized.
The correct way is to start with a computer that have all updates installed and rebooted to minimize them skewing the results.

The computer is connected directly to the internet/cable modem if possible and a test is run, if it is slowed down then the issue is with the cable modem or the ISPs network.

If not then the router is connected and a test is run again.

If this test is slowed down then the error is either as mentioned above or with the router, so start again with only the cable modem/directly connected and see and then test with the router again. Do this a couple of times.

If the test behind the router is not slowed down, then connect the switch and connect the computer to the switch and run the test again.
If the test is slowed down, then the error can be in the internet connection/cable modem, router or the switch, so start again from the cable modem/directly connected and work your way down again to see if the slow down always occur at the switch or if it occur earlier.

If it does not slow down behind the switch, then you have either not found the error and need to keep trying or you have a problem somewhere else, like on the server, but do not just base your decision on a single or a few runs. Run many times right after each other and also run with bigger spaces in between.

An error found is a success here and you can limit the search. No errors found can only be used as an indication when you have MANY such results!

I am about 99.9% sure if everything is hardwired, it is your HA box. I know it is my Raspberry PI 4B but, I am OK with it. It’s job is not to sit there and measure speed. I basically use it as a sensor just to indicate there might be a problem. If I have doubts as far as my circuit getting the speeds it needs to be, I use my computer to spot check.

I do not know if anyone has a Pi 5 to check it. It certainly would be interesting to see the results. It might convince me to go down that path…

I’m having same issue. I have 5Gb down with 2Gb up fiber. HA is running on VM; my windows server and Arista NFGW also running on same VMhost. Running speedtest on windows server via http I get about 3Gb down and 2Gb up (limit of speedtest web) and 5Gb down, 2Gb up with speedtest client. HA reporting 1G down and 500Mb up. Both VMguest use same 10G physical NIC. It’s not my hardware. I tried increasing the vCPU from 2 to 4 and mem to 16; no change. My windows server is using 2 vCPU and 16megs. Made sure both windows and HA test was pointing to same Speedtest server