And what events on the PC signalled the start and stop time? Was it started the second the mouse button was pressed or was it started when a dialogue showing progress appeared? If it was the latter then it may be a virus checker or browser cached a little before the dialogue got around to appearing.

Hmm, I'm beginning to wish I'd never started this! I just timed it using an ordinary watch at 2 minutes, 4 seconds. Absolute accuracy is not important as a second either way wouldn't have made any difference to the conclusion. I realise now that what must be happening is that the browser is caching some of the download before the point at which you tell it where to save the file - I just tried it again and did it in 1 minute 30 seconds, which is even more impossible.

Edited by kasg (Thu 13-Mar-14 15:24:48)

It had to be something in the timing was all I was trying to diagnose .

Yeap browser start download before you have picked the save location, hence why you sometimes see massive burst speeds if you are slow picking the save location.

Yes, as soon as you click a link in it starts to download in the background whilst waiting for you to say where to save it. That is one of the reasons someone suggested running TBBMeter as you would see when the actual download started.

In reply to a post by ian72:

That is one of the reasons someone suggested running TBBMeter as you would see when the actual download started.

or a command line tool such as wget.

Thanks to you and ian72, I should have worked that out earlier.

In reply to a post by mr_mojo:

http://www.thinkbroadband.com/speedtest/results.html...

Seems to be that Chrome maxes out a CPU core on that test result, which may be causing it to peak at 500meg. Not sure.

Chrome lags because of recent problems with Pepper Flash 12.0.r0, causing high cpu useage.

I have used Internet Explorer for running speed tests for a long time and Chrome for general use, as IE performs more reliably in general on multi-threaded speed tests than Chrome, due to various bugs in Pepper Flash.

Some things that can also slow your system down when you run a browser based speed test:
- The browser cache writes out files to make room for your browser test data, even though it's just in memory. There can be quite a lot of files if they are small images for example.
- Any files written may need to be virus scanned, if you don't have that turned off for those files.
- Any files written may need to be indexed by windows indexing system, which requires writing additional data.
- Windows pre-loads/pre-fetches files and can do this when you don't want it to.

If you run a resource and performance monitor while you run your speed test you can see what exactly is causing the high load:
http://en.wikipedia.org/wiki/Resource_Monitor

Of course if you're running Linux then most of this advice will need to be modified, but still may be useful.

In reply to a post by MUD_Wizard:

In reply to a post by mr_mojo:

http://www.thinkbroadband.com/speedtest/results.html...

Seems to be that Chrome maxes out a CPU core on that test result, which may be causing it to peak at 500meg. Not sure.

Chrome lags because of recent problems with Pepper Flash 12.0.r0, causing high cpu useage.

I have used Internet Explorer for running speed tests for a long time and Chrome for general use, as IE performs more reliably in general on multi-threaded speed tests than Chrome, due to various bugs in Pepper Flash.

I seem to get better speeds in IE, not noticed this before.

http://www.thinkbroadband.com/speedtest/button/13948...

Though the TBB speed tester tends to crash in IE for me at these sorts of speeds. The ISPA one works.

In reply to a post by mr_mojo:

No, my point is that it is ridiculous that so many people say that FTTC is a 'joke' and we should be spending £30bn+ of taxpayer money subsidising FTTP rollout to everyone. FTTC is more than enough for right now.

I'm almost certainly going to downgrade to the £10/month 100/100 package next month - it's absolutely pointless having over 100megabit/sec right now.

I tried a 10 thread download to MSDN to download a windows server ISO and it is capped at 100mbit/sec.

BT has made a very wise decision not to go past FTTC right now.

PS: It's great that I get this speed (and also have BT FTTP as an option if I wanted), so I'm not being ungrateful. I just wanted to prove that this sort of speed is so silly. It reminds me of when everyone had 56k modems and some people were using dodgy firmware to get 10 or 20mbit/sec on NTL cable modems back in the day. Great - but you'll quickly realise that being 100x faster than the normal broadband speed means your held back by infrastructure and the internet as a whole.

If everyone received a perfect service on FTTC and got full 80mbit connection speeds, then yes FTTP would be useless.

As it is now, with people living 500, 1000M+ from the cabinet, over a mixture of 60, 70 year old copper/aluminium, the service cannot be called futureproof at all, or even adequate for some, if they are far enough away.

The main benefits of FTTP which make it such a common sense investment for the future are:

NO interference, line length issues, no CRC errors, interleaving
Can be upgraded over the next few decades to provide more than enough bandwidth by just changing the hardware on each end of the fibre.