Thinkbroadband and Speednet

Hi

I have used Thinkbroadband for years to regularly check my speeds, always with NTL and later Virgin. Since moving to a non-cable area, my speeds have fluctuated wildly but never going over a download speed of 5mb.

I eventually phoned Virgin today, and they asked me to record speeds over a 24 hour period, but to use Speedtest.net, who I had never heard off.

Using them to check my speeds, using a line, not wireless, the speed are between 75% and 100% faster than those recorded via Thinkbroadband. The Thinkbroadband tests take longer so must be more thorough, but why is there such a difference. Its most annoying as Virgin will now say I don't have a major problem; when I know I do.

In reply to a post by macshimi:

Using them to check my speeds, using a line, not wireless, the speed are between 75% and 100% faster than those recorded via Thinkbroadband.

Download, upload or both speeds?

The better way to record speeds is to use the TBB, Speedtest.net , BT (not in your case) and VisualWare. Record the results but also take a screen shot of Task Manager which give a graphical representation of all three/four. It is easy to add a line as the where your speed should be and then to see where it is not achieving it.

Andrue, it is the Download speeds that differ; the upload are the same.

Thanks, MHC; for Virgin I have to use Speedtest.net. I want to find out why tests, one on Speedtest and the other on Think BroadBand, and done at the same time vary so much.

Have a read of the threads in TTTS ... there is a lot of information there about the various speedtest sites. In brief, speedtest.net throws away the slowest samples and a few high ones which effectively biases the result upwards.

Speedtest.net also uses multiple connections (4 last I checked) which helps overcome the kind of congestion manifest on VM's network but is possibly less effective on BT (Retail)'s.

In reply to a post by qasdfdsaq:

Speedtest.net also uses multiple connections (4 last I checked) which helps overcome the kind of congestion manifest on VM's network but is possibly less effective on BT (Retail)'s.

Any idea why speedtest.net seems to speed up at the end? Is that the effect of it throwing away slower results and replacing them with faster ones over time? When I run it the download test usually jumps straight to 50Mb/s then in the final two seconds climbs higher sometimes getting up to 58Mb/s. The upload doesn't do the same thing and just rises fairly quickly to 14Mb/s.

I've never seen it speed up at the end.

However I've been told it "bins" the test into several different sections, and takes the average of the fastest section. Windowed sample, not average of the whole test. Basically what MHC already said.

In reply to a post by qasdfdsaq:

I've never seen it speed up at the end.

However I've been told it "bins" the test into several different sections, and takes the average of the fastest section.

More detail.

Beat me to it ...

Thanks.

There's something else they implemented recently (well about a year or two ago) to counter the "Kaspersky effect" and isn't mentioned in that documentation though. Not sure exactly what is done.

speedtest.net to preston server gets my max speed of 106 Mbits 9/10 tests. That is my suggestion of which server to use, and speedtest.net stores your results so you can go back and show VM(who im with) the results,

I also suggest if you dont already have one, set up a TBB quality monitor for your IP address.this can outline high ping and jjitter which will affect the maximum speed you will get

I hope this helps

The Preston server is hosted on VM's network, so is OK for testing your line speed, but useless to test your actual internet connection.

Ironically it used to give ~15mbps to VM customers and ~75mbps to BT customers (and >90mbps to France).

Glad to see that's fixed...

In reply to a post by qasdfdsaq:

to counter the "Kaspersky effect"

No idea about that I'm afraid... I use Macs.