Can your OS affect download speeds?

I have been trying to help an elderly neighbour who got left out of the shambles of the AIO cabinet migration recently done in our village.

Like everyone else who was and still is on ADSL the performance dropped dramatically after the FTTC was activated.

He has a windows 10 laptop and the connection was almost too slow to do any speed testing so tried my Windows 7 one. We did in the end about a half a dozen tests on each laptop, spread between thinkbroadband and bt wholesale checkers. All the speeds recorded on his laptop were between 0.15Mbps and 0.35Mpbs and on mine between 0.4 and 0.7.

While it is not impossible that on 12 random tests of an erratic service the top 6 would be on one machine I think the probability is so low it can be discounted.

My laptop was connected by wireless and his on most of the tests connected by ethernet, and rather strangely all my results were higher than the line profile of 0.31Mpbs

The only other factor that might influence the results was the upload speed on both devices and all tests was a consistently all most non-existent 0.03Mpbs and I wondered if Windows 10, Microsoft Edge and the other Windows 10 based software were less tolerant of this than Windows 7 and Firefox.

You need to repeat te tests using WIRED for both PCs.

However, I doubt if you will find any significant difference once you do that.

What was his ADSL speed before the change and what does the BT DSL checker say?

We report to 0.1 Mbps resolution on the figures, which needs to be bourne in mind. Data stored to full resolution, so graph will show the actual recorded speeds.

30 Kbps upload suggests a problem, and you need to look at the connection speed of the ADSL modem, it may be that it just needs a restart.

Also if person is on the old IPStream Max (ADSL) service and ADSL2+ is available getting them migrated to that or a LLU operator may give better performance.

I could not set my laptop to a wired connection as the BT Hub required permission to do so, and the smart set up which apparently you need to allow this was not showing on the hub manager.

But as you say it will have made little difference and certainly it made none when testing on my neighbours device.

It is troubling, of course, to see you achieving speeds that ought not be possible, given the IP Profile. Assuming, of course, that the IP Profile reflects the actual sync speed ... and that resyncs weren't occurring regularly.

The next doubt would be that the wireless connection wasn't actually to the hub you expected it to be, and was instead connecting elsewhere.

After that, you start to doubt the way the speedtester is measuring, and whether it is affected by anti-virus software. I would step away to manual, command-line testing at that point.

Without seeing the actual speed tests its hard to say, rounding and bursty wireless can mean you get oddities, particularly on a line that may be seeing errors and re-transmits.

In short without more info i.e. at least a link to a speed test and the actual line stats difficult to say more other than their line is slow.

Andrew

I rebooted the BT hub - I presume the modem is internal to that.

I did not want to complicate the story but previously the line was with Talk Talk and he was getting 1Mbps and moving to BT and so away from an LLU service after the AOI cabinet went live last October may account for some of the drop. His contract with BT gives a range of 0.4 -1Mbps with a guaranteed speed of zero.

As an aside you do have to questions the ethics of a company that sells a zero guarantee service to an elderly person although I understand when I put a complaint in about this that zero actually means 256K

He is on ADSL2.

Yes I have certainly has speed tests in the past which have been higher than the line profile. At very slow speeds I suppose bursts have more impact than at higher speeds.

In reply to a post by gerarda:

He is on ADSL2.

Do you mean that? As distinct from ADSL2+?

I missed the + out . was just meaning to say he is not on ADSL Max

OK.

This won't affect the hassles you are having with speedtesters, but it can be worth forcing ADSL2 mode instead of ADSL2+ - at least if that modem/router can be told to do that.

ADSL2 sometimes works better than either ADSL or ADSL2+ on longer lines; it did for my mum.

Bursts are more of an issue at low speeds, hence why the graph is important

There still remains the basic question - is it just a coincidence that the top six recorded speeds were on Windows 7 and the bottom 6 on Windows 10?

Without access to all the information - which starts with the speed test results, i.e. the unique ID so can look at raw data and the actual connection speed data from the modem, the reality is impossible to say.

If this was the same PC running Windows 7 and then Windows 10 then you might be able to make comparisons, but with two different sets of hardware you introduce even more multiple variables.

In short cannot answer the question with the information I have.

I suspect there is not much data anyway. I did not keep a record but I think only a couple of tests were done using your checker on the Windows 10 machine. Downloading the rest of the page content made it much harder to do the test than on the BT wholesale checker. but the postcode we supplied was IP14 3RL and the tests were between 5 and 7 pm last Friday.

Using flash http://tbb.st/148433208272590482692 and the other http://tbb.st/1484333038838371955 first windows 7 and second windows 10

The stand out is the latency which is worse than a satellite connection and is what I'd be looking to check if this is the case in reality

Edited by MrSaffron (Sun 15-Jan-17 17:09:03)

Andrew

That's very helpful thank you.

If one of those was from Windows 7 then it blows the OS issue, It must have been one of the tests we thought had not completed as it took so long to do.

Unless we can persuade to migrate him onto the AOI cabinet in a reasonable timescale so he can upgrade to FTTC it looks like its worth getting an engineer out to check out the latency.

I think the latency is probably as bad as recorded as that would explain why it acts much slower than even the slow speeds would suggest it should.

I do wonder if the AIO cabinet is somehow interfering with his line even if the database says he has not been moved onto it.

You might want to remove that coma at the end of that second result link

Paul

And the ADSL stats would give some idea of if the line is under performing or seeing lots of errors