TBB speedtest anomalies

I ran a speedtest this morning and found vastly different speeds on the two parts of the download test. So I then resynced my router from 6486 to a speed of 6621Mbps and then run a speedtest and I am getting strange results.
http://www.thinkbroadband.com/speedtest/results.html...

Only 1.4 on the first test and then the expected 5.5 on the multithread test.
What is going on here ?
I am with Plusnet and their current line speed is 5.6. So I can see no reason for this artificial capping. The line is and has been stable but seems to be running slowly this morning.

Routerstats are:
Line sync = 6621
Atenuation = 53.5
SNR = 3.6db

I had similar differences once because of an internal network problem: a mismatch between my HH3 gigabyte port and my cheap TPLink gigabyte switch. As with yours, it only showed up on TTBx1 test.

All was fine when I used one of the router's 100Mbps port (and still get gigabit between devices on the switch). No other issues with the switch.

In reply to a post by StephenTodd:

I had similar differences once because of an internal network problem: a mismatch between my HH3 gigabyte port and my cheap TPLink gigabyte switch. As with yours, it only showed up on TTBx1 test.

All was fine when I used one of the router's 100Mbps port (and still get gigabit between devices on the switch). No other issues with the switch.

Interesting, I too have a gigabit local network accessing the internet via a 100Mb port on my adsl router and all PCs on the gigabit network show the slow speed cap.
Accordingly I plugged the PC directly into the adsl router and there is no change. It is still too slow on the first part of the TBB speedtest and that is on either network speed.
Ideas anyone why the speed seems capped.

Another slight possibility is an MTU mismatch. Not too common on modern OS's as they usually adjust automatically.
That often hit single threaded tests (such as TTBx1) more than mutlithreaded ones (such as TTBx6).

(for info see http://www.cnet.com/uk/news/tip-increase-internet-sp... , an old post as it is mostly an old problem).
There used to be a free program called MTUSpeed that would help check; I ca only see a paid MTUSpeed Pro around now.

Edited by StephenTodd (Thu 04-Jun-15 08:18:49)

In reply to a post by StephenTodd:

Another slight possibility is an MTU mismatch. Not too common on modern OS's as they usually adjust automatically.
That often hit single threaded tests (such as TTBx1) more than mutlithreaded ones (such as TTBx6).

(for info see http://www.cnet.com/uk/news/tip-increase-internet-sp... an old post as it is mostly an old problem).
There used to be a free program called MTUSpeed that would help check; I ca only see a paid MTUSpeed Pro around now.

Thanks for your thoughts Stephen.
While trying out the connection via the slower lan I ran a small tool called TCPOptimiser. This suggests the optimum settings for the current conection speed (receive window,MTU etc) I ran the calculator to find the largest MTU without fragmenting. It was all spot on however. So I don't think that is where the issue lies.
I will give it a try on my Win10 test machine. to rule out an OS issue.
BTW that link you gave me is dead now.

Not dead, just wrong http://www.cnet.com/uk/news/tip-increase-internet-sp...

Sorry about the wrong link: fixed. I forget that the forum software corrupts links when immediately followed by punctuation.
I'm not surprised the MTU was not the issue, but it was worth a try just in case.
Good luck finding the problem. When I had the issue it did not seem to affect real browsing or downloads/uploads too much, just the TBBx1 test. I hope that is so for you too.

I was seeing the same sort of thing at much lower speeds (we are a good hike from the cabinet so only get about 16Mb down and 1.5Mb up). The single threaded part of the TBB speedtest was often reading much lower than the synced speed (has been down as low as 1Mb at busy times).

Changed from the BT HH5 to a Billion NL8800 and both tests now show the full speed. May just be coincidence or it may be that some routers get indigestion using the speedtest.

Well without my changing anything the speedtest went back to normal after the Plusnet 'Current line speed' was reset a couple of days after a reboot and sync at a slightly higher speed this was around June 5th.

Then on Friday 12th we had a power cut and the connection speed and Plusnet Current line speed dropped due to the time of day. After a reboot on Saturday Plusnet eventually reset the Current line speed this morning to the correct level for my sync speed and I am now back with the low speed 1.43Mbps on the single thread speed test !!

I really don't like Plusnet gerrymandering with the ip profile and download speed. Why can't they just leave things to the BT ip profile which is always higher than Plusnet will allow anyway (BT = 6.12, Plusnet = 5.7).
I am finding Pulse8's offering more attractive each time this happens especially when I get crippled using a single thread download.

I don't think the 1.43Mbps is anything to do with the Current line speed. It is something else strange in parts of Plusnet's network. I often have the same effect, but it doesn't seem to affect live working.

I would expect a higher than 5.7Mpps setting if your IP Profile is 6.12Mbps. More like 6.1.

Limiting your output speed from an ISP's servers is actually quite common, if not ubiquitous, though few talk about it. If I'm not mistaken both Zen and AAISP mention it for example. The reason is that your stream is then buffered within the ISP's network rather than suffering huge packet loss when it hits the MSAN/DSLAM at full speed.