Single Thread Speed Tests

I've been looking for something to ratify the single thread results that I am typically seeing for my Pulse8 76/19 connection:

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

I have been using testmy.net's single thread automated test for comparison and have been getting results between 50Mbps and 72 mbps with an average of significantly over 60 Mbps.

Whilst searching for singe thread tests I also came across a TBB single thread tester, which gave results of over 70Mbps

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

No idea of the vintage of the TBB single thread tester or whether it was ever released, but it does give results of that are significantly different to the TBB Flash and non-Flash single/multi thread tests.

The question is: which set of data should I believe?

Both the testmy.net and the TBB single thread tests suggest that I shouldn't worry too much about the performance of my line, however the normal TBB test does suggest there is an issue worthy of further investigation.

I would welcome your thoughts.

The version you found was http://labs.thinkbroadband.com/speedtest/?site=omega...

BUT due to on-going development there was a bug (literally fixed it once I saw your post) where it was actually just doing one download sequence but using six threads - indicator was exact same lines for tbbx1 and httpx6 test

So refresh the page and run the test again, the behaviour will be the same as our normal version now i.e. http://labs.thinkbroadband.com/speedtest

The version you expressed a worry about is the old flash version (not in development) so run the fixed version linked to above and can then comment. NOTE: The flash is using a socket that is NOT HTTP, so with some providers there can be a difference due to traffic management between the latest versions and the old flash.

You are correct; the omega link is the one I used earlier

The only significant difference between the two tests now is the introductory blurb that says Welcome to the Single Thread tester before running both single and multi thread tests! Is some more code tweaking required?

'Single' thread result: http://www.thinkbroadband.com/speedtest/results.html...

Multi thread result: http://www.thinkbroadband.com/speedtest/results.html...

Neither provide an upload graph.

Latest testmy.net result @ 16:20 is 68.47 Mbps.

One further question if I may. If you are pushing the non-flash version, why does the speed test default to the Flash version when you are logged in?

The omega one shows an upload graph for my connection, it seems that the speed checker glitched or my downstream got 2.5Gbps for a very very short time and then settled down to a bit of congestion
TBB Speedtest

Paul

Not sure why no upload graph, will try and remember to look in the morning, its not a bug I've introduced but may be something odd in your data, will have to grab the raw data and see whats wrong.

If someone has flash installed we default to that so it will link to their profile, and also running both concurrently allows us to compare the two versions across a large userbase.

No idea why 14ms in you got a figure of 2578.14 Mbps, but as we do guard against these oddities while plotted it won't have affected your average.

In reply to a post by MrSaffron:

No idea why 14ms in you got a figure of 2578.14 Mbps, but as we do guard against these oddities while plotted it won't have affected your average.

Totally agree.

I have seen it happen a few times now at the start, I always put it down to I got the full 2.5Gbps (max bandwidth) of the fibre cable and then BT's system went Whoa slow down dude, where it then went back to the speed I am suppose to get.

Maybe its just a coincidence that that speed of 2.578 Gbps matches the 2.5 Gbps that BT say they use per fibre.

I could be wrong though.

Paul

It might be possible but also the Ethernet to the PC would need to support those speeds for that to be the case.

Techniques present in the flash to mitigate these spikes are not in the HTML5 as wanting to see all the quirks to ensure that real stuff is not mitigated by mistake

In reply to a post by MrSaffron:

It might be possible but also the Ethernet to the PC would need to support those speeds for that to be the case.

Well our house is all networked up with a 1Gb LAN and all the mobile devices using Wi-Fi.
My PC is using a fast 1TB SSD capable of over 2.5Gb speeds.
The only issue here would be our wired LAN being only 1 Gb, so maybe it was only a glitch or BT breaking for a second or maybe our LAN saying Whooooa slow down there.
But saying that, it couldn't of been our LAN capping it due to it would of been capped to about 800 to 900 Mbps where it wasn't, so I am still leaning towards a glitch or a BT issue.

I am tempted to upgrade to 10Gb Switches so that the downloading doesn't take too much up on the LAN, I already have CAT6 and CAT6A cable installed.
I know some of the cables were double screened along with a single wire down the middle, that was the expensive cable.

In reply to a post by MrSaffron:

Techniques present in the flash to mitigate these spikes are not in the HTML5 as wanting to see all the quirks to ensure that real stuff is not mitigated by mistake

Its probably just a glitch, but for it to happen a few times now at certain times make me wonder.
I know speedtest.net is also having some strange issue with our connection where it would sit there for about 10 seconds then start giving us a 0 ms latency and completing the downstream test very fast, it didn't give the 2.5Gb speed, but it still had an issue.

Paul

In reply to a post by MrSaffron:

Not sure why no upload graph, will try and remember to look in the morning, its not a bug I've introduced but may be something odd in your data, will have to grab the raw data and see whats wrong.

Did you manage to have a look?