In reply to a post by MrSaffron:

Its interesting in that the new FTTP areas with Project Lightning are done minus telephone line.

Should have telco in those areas in the not too distant.

I'm making no assumptions; TBB works fine on other ISPs both myself and friends have access to and the test itself has no inherent issues with single thread performance I'm aware of. TBB staff monitor test results and you can see published results of the tests in their monthly reports.

But don't just take the results of one test - try basic web browser file downloads, FTP transfers, other speed tests that do check single thread performance (many don't). Examples include speedof.me and dslreports if you manually set the number of threads. I've repeatedly found TBB's results to be fully representative of download performance.

Steam, Origin, etc use many download threads so bypass the problem.

Just for comparison, here are results from my own BT and Virgin connections:
http://www.thinkbroadband.com/speedtest/results.html...
http://www.thinkbroadband.com/speedtest/results.html...

But like you say you could be free (or mostly free) of the problem - some do get solid performance on VM which is why it's such a strange issue. But that VM test above is supposed to be a 200Mb connection on Virgin...

Edit: Just realised the BT test was from after midnight so not a like-for-like comparison. Here's another from peak time to show it's not a cherry-picked result: http://www.thinkbroadband.com/speedtest/results.html...

Edited by deleted (Mon 14-Nov-16 22:38:58)

In reply to a post by watercooled:

But don't just take the results of one test - try basic web browser file downloads, FTP transfers, other speed tests that do check single thread performance (many don't). Examples include speedof.me and dslreports if you manually set the number of threads.

Yep, that's what I said, I use many different mechanisms to transfer files from game downloads and to/from my office and other locations. Every one gives me a better result that TBB reports. I also use FileZilla a lot for secure ftp, it does multi-threaded but reports the speed for each thread. I have no idea why there is such a discrepancy though, just wanted to point out that TBB are not infallible.

There was a very informative exchange between Virgin Media and Ofcom over testing of the 200 Mbps service back in 2015, the summary being the three threads used by the SamKnows system was not testing the full capability of the 200 Mbps service and that more threads were going to be added.

Of course tbb is not infallible but then no company is infallible, as one of the few to offer a single http download test with no accelerators we will often be going against the grain.

filezilla = threaded
steam = extremely threaded (typical game download for me is at least 20 connections to steam servers)

@Daemon66: Of course, it is wise to use more than just one test when you're trying to diagnose a problem. I just provided TBB as an example, but in my case it matched up with pretty much every other test/download I tried. I'm not sure why it would be an outlier in your case, but there are a load of reasons from the wider network to something going on with flash player or security software for example.

@MrSaffron: I'd argue that even using three threads is not completely representative of real-world download speeds, let alone more! They should at least provide a single threaded test separately and explain the difference to readers IMO. I did provide Samknows with some feedback on their testing methodology and how it can mask issues such as this. There's no use having 200Mbps available with 50 threads if, like me, you get <9Mbps on standard file downloads so even 500MB GPU drivers end up taking closer to 10 minutes rather than the 20 seconds or so you'd expect with 200 meg! It's really not nitpicking is it?

I don't suppose you have a link to that dialogue between Samknows and Virgin (assuming it's public)?

@Chrysalis: Yeah, I found the same with Steam, speedtest, and the like, so at least huge game downloads were quick (although even they started struggling to get over 150 just before I left as it was getting so bad) but unfortunately things like equally-large PS4 downloads and Windows Update/Store were painfully slow, with games taking upwards of 12 hours for example - quite a departure from the expected ~40 minutes!

I have a few friends/relatives in the same area still on VM and have been able to test their connections. According to hostnames, they're all on the same CMTS chassis but different line cards, and all are affected just as severely. I just wish I could get somewhere with VM on the issue!

Also, sorry for going off-topic on your thread, hope you don't mind discussing it!

Edited by deleted (Tue 15-Nov-16 19:26:24)

The snippet is buried somewhere in the Ofcom 2015 speed test analysis

In reply to a post by watercooled:

@MrSaffron: I'd argue that even using three threads is not completely representative of real-world download speeds, let alone more! They should at least provide a single threaded test separately and explain the difference to readers IMO. I did provide Samknows with some feedback on their testing methodology and how it can mask issues such as this. There's no use having 200Mbps available with 50 threads if, like me, you get <9Mbps on standard file downloads so even 500MB GPU drivers end up taking closer to 10 minutes rather than the 20 seconds or so you'd expect with 200 meg! It's really not nitpicking is it?

I don't suppose you have a link to that dialogue between Samknows and Virgin (assuming it's public)?

@Chrysalis: Yeah, I found the same with Steam, speedtest, and the like, so at least huge game downloads were quick (although even they started struggling to get over 150 just before I left as it was getting so bad) but unfortunately things like equally-large PS4 downloads and Windows Update/Store were painfully slow, with games taking upwards of 12 hours for example - quite a departure from the expected ~40 minutes!

I have a few friends/relatives in the same area still on VM and have been able to test their connections. According to hostnames, they're all on the same CMTS chassis but different line cards, and all are affected just as severely. I just wish I could get somewhere with VM on the issue!

A couple more examples of 'real world' download speed using a single TCP/IP socket.

Operating system updates. A recent IOS update for an iPhone of mine took 40 minutes for the download - Only VM comment from the guy who rang me after I cancelled - "ouch!"

The Sky box downloads the content onto the internal HD. The playback commonly catches up causing the box to indicate 'end of recording'.

I think sony psn downloads are single threaded, but not entirely sure.

windows update is definitely single threaded.

so those may give you a clue if tbb is accurately reporting single threaded performance.

In reply to a post by Chrysalis:

filezilla = threaded
steam = extremely threaded (typical game download for me is at least 20 connections to steam servers)

Yes but FileZilla reports the transfer rate per thread so the single thread speed IS shown and it is generally higher than what TBB reports.