[kwikbreaks]

Going back to the original post...

When I was on 50Mbps before VM borked my connection by oversubscription...

I found that the Speedtest.net London and Birmingham servers would consistently show 50Mbps. That was true even with some short term drops in that speed as shown by an external monitoring program. That's because ST reports peak speeds rather than the average. Most of the time on my connection back in the days when it worked usually ran close to peak throughout the test anyway.

ThinkBroadBand never came close to 50Mbps. It reports average speeds but even the peaks were showing less than ST. This was absolutely consistent. Some (even on VM) get good results from TBB so this must in some way be down to routing but it was always the same for me.

IMO the BBMax speedtest uses far too little data to be reliable on high speed connections. My guess is that they keep the testfile size low because their server and connection aren't fast enough to handle higher volumes. I reckon low speed tests from them reflect more on their tester than your own connection. This could have changed as it's been a long time since I used their test.

Additionally many AV products impact the results reported by speedtesters. You really should be using other software as a crosscheck on what their tests report. Something like netmeter is good enough although that too will see the data after the AV so may be affected by AV buffering and scanning. I use the built in monitoring in the Tomato firmware on my router - that isn't immune to AV either though as it will only deliver what the PC can take. If you suspect your AV then turn it off for the duration of the testing.