Reliable speed test website?

[StevieP]

I finally had my 76MB fibre connection installed yesterday and wondered what the most reliable website was to gauge the actual speed? (when I signed up the Plusnet website said I could expect to get roughly 72MB)

I've tried www.speedtest.net but the results seem to vary wildly depending upon which server I choose!?!? (ranging from the low-20's up to the mid 70's)

I added a torrent to my torrent client and it seemed to max out at around 7.6MB/sec - given that I used to get roughly 1.4MB/sec on my circa 14MB ADSL connection, I'm assuming it's pretty much 76MB if you pro-rata things up?? (or isn't that how things work??)

Thanks for any pointers....

 

[Luke_Bowman]

ISP's market their speed in 'bits' not 'bytes'.

Bytes are what normal files are measured in and 1byte = 8bits.

So if you divide your speed by 8 then that's your actual download speed you should see on Steam/uTorrent/Chrome.

7.6*8 = 60.8mb/s so that seems pretty good for a torrent

 

[asim18]

I've tried www.speedtest.net but the results seem to vary wildly depending upon which server I choose!?!? (ranging from the low-20's up to the mid 70's)

Yeah this is what most people forget when testing broadband. You will get full speed to servers in "well connected" data centres, but lots of data centres still don't have good connectivity/peering arrangements with certain ISPs.

For a good ISP, You should really be getting maximum throughput right across the UK and major parts of Europe.

I get a solid 150Mb/s to all major locations around Europe and 140Mb/s to New York. Sometimes I get faster speeds to Scotland than I do to London servers so you know the network has good routes in place for its traffic.

I think speedtest.net is the best really, you cant beat their choice of locations.

 

[Glanza]

Speedtest is the go to.

 

[Steveocee]

Ask your ISP for their recommended site.

BT generally ask for you to use the BT wholesale speed tester.
Speedtest.net is a good "general" site, just try to make sure it tests from the same geographical server.

 

[asim18]

BT generally ask for you to use the BT wholesale speed tester.
Speedtest.net is a good "general" site, just try to make sure it tests from the same geographical server.

They do this because they want you to test the speed to "BT" instead of the actual "internet". "The internet" is not located at BT's wholesale speed tester.

OP wants to gauge his actual internet performance. The only way to do this is to test various global servers which take various routes across the internets.

 

[bledd]

Make sure you're doing speedtests over ethernet only. Doing them over wifi/powerline is pointless.

 

[bitsNbobs]

http://www.mybroadbandspeed.co.uk/

Nearly always gives me accurate results.

Speedtest.net like you, varies quite a bit. Some servers which used to be accurate for me now report around 5Mb less on the upstream and anything upto 15Mb less on the downstream than what my actual connection goes at. Using that site i have to try various servers within about a 50 Miles radius to find one that reports accurately.

 

[asim18]

Amazed to see the lack of understanding of what the internet is.

There's no such thing as "what my actual connection goes at". The throughput you attain between two computers on the internet is route dependant and the internet consists of millions of routes. Because you're getting 15Mb less using a particular server/datacentre/route means you are actually getting 15Mb less, it's not some bug in the measurement which is causing inaccuracy. The fact that you can squeeze 15Mb more by using a better route is dependant on that specific route.

 

[Chugz]

Just use speedtest.net and take 3 tests on the same server. Average out the results, and hey presto, you've got your speed.

 

[Steveocee]

They do this because they want you to test the speed to "BT" instead of the actual "internet". "The internet" is not located at BT's wholesale speed tester.

OP wants to gauge his actual internet performance. The only way to do this is to test various global servers which take various routes across the internets.

Are you joking?

BT are responsible for the speed between the customer and the breakout to the internet. Why would they be responsible for the speed of the route to some random joes website the other side of the world?

If you are a BT customer and you test to BT's site, you are testing to BT's core, that is the best place you could test to as it will give the truest representation of actual throughput available on that line.

Of course they have an obligation shoving it out to the internet but if there was a problem there then they'd have WAY bigger problems than a couple of people saying they are getting varying speeds.

Practical example, I am on an uncontended guaranteed 25Mb service;
I speedtest from myispspeedtest.com (which IS at the core of the ISP network and is internet facing so is technically "the internet" and get 25Mb
I speedtest from Ookla in London and get 16Mb
I speedtest from Ookla in Germany and get 10Mb

Should I be reporting this to my ISP? No. I shouldn't because my connection speed to the breakout point is exactly as promised. Why would my ISP be responsible for speeds once the connection is handed off?

 

[GAM8LE]

Speedtest.net is my go to. But I tend to do three runs and take an average.

I did a comparison of a few before and got basically the same results.

 

[asim18]

Are you joking?

BT are responsible for the speed between the customer and the breakout to the internet. Why would they be responsible for the speed of the route to some random joes website the other side of the world?

If you are a BT customer and you test to BT's site, you are testing to BT's core, that is the best place you could test to as it will give the truest representation of actual throughput available on that line.

Of course they have an obligation shoving it out to the internet but if there was a problem there then they'd have WAY bigger problems than a couple of people saying they are getting varying speeds.

Practical example, I am on an uncontended guaranteed 25Mb service;
I speedtest from myispspeedtest.com (which IS at the core of the ISP network and is internet facing so is technically "the internet" and get 25Mb
I speedtest from Ookla in London and get 16Mb
I speedtest from Ookla in Germany and get 10Mb

Should I be reporting this to my ISP? No. I shouldn't because my connection speed to the breakout point is exactly as promised. Why would my ISP be responsible for speeds once the connection is handed off?

Were you supposed to quote something else? I cant see anywhere where I've talked about responsibility or anything lying with anyone as such.

I'm just pointing out the fact that the internet only starts at your nearest "hop". Just run a trace route to any popular website if you don't believe me see how many hops are external. This is where the external peering arrangements come into place and how much your ISP is willing to pay for this connectivity, and this is why ISPs run caching servers for the major internet corps too.

If you still don't believe me just look at how many people complain about Hyperoptic's 1Gb/s service being contended as hell, yet their speedtests to the nearest hop max out to full speed!

 

[andy_mk3]

I always get better results on speedtest.uk.net, much better site too.

 

[bitsNbobs]

Amazed to see the lack of understanding of what the internet is.

There's no such thing as "what my actual connection goes at".

So you no connection has limits on the speed it can go? Thats interesting ill remember that highly amusing pearl in future.

The throughput you attain between two computers on the internet is route dependant and the internet consists of millions of routes.

Its either dependant on "throughput you attain between two computers" (oh i can use quotation marks also). Or its dependant on "route dependant and the internet consists of millions of routes" Please decide!!
Id be worried if any connection attempt i make has to make a "million" routes first though.

Because you're getting 15Mb less using a particular server/datacentre/route means you are actually getting 15Mb less,

No it doesn't, when you connect to the net your modem authenticates, goes through a hand shake process. Tones and how good or bad they are then determine what speed the connection syncs at. When i get 15Mb less than what my connection syncs at or back to "what my connection goes at" its because the speedtest server is incapable of sending a higher amount of data or Mbps, than what i am capable of receiving. It would be no different to if i was P2P'ing files with you and i had a connection capable of 100Mb download but your upload is only 10Mb, meaning i would only be able to download at 10Mb, NOTHING to do with my connection, NOTHING TO DO with the route, its down to HOW fast YOUR END is. (or in this case the speedtest server).

it's not some bug in the measurement which is causing inaccuracy. The fact that you can squeeze 15Mb more by using a better route is dependant on that specific route.

You seem to be confusing latency with throughput. Or maybe even contention which are different matters entirely.

Practical example, I am on an uncontended guaranteed 25Mb service;
I speedtest from myispspeedtest.com (which IS at the core of the ISP network and is internet facing so is technically "the internet" and get 25Mb
I speedtest from Ookla in London and get 16Mb
I speedtest from Ookla in Germany and get 10Mb

Should I be reporting this to my ISP? No. I shouldn't because my connection speed to the breakout point is exactly as promised. Why would my ISP be responsible for speeds once the connection is handed off?

Exactly and a good example on how speedtest servers vary. Some will measure closer to what your 25Mb connection is capable of and others and well and truely off.

 

[Rainmaker]

The best I have found is DSLReports' speed test. It is HTML5 and easily maxes my 152Mbps VM line even over wireless N. Like this:

The ThinkBroadband Flash tester is also quite accurate, and they have some other nice tools available too.

EDIT: A bit cheeky and away from its intended purpose, but the BT Wholesale speed tester also seems to give reliable results even for cable customers. Just click 'yes' to agree that you have set up your 'ADSL' line and then click to test. Voila!

 

[Steveocee]

Were you supposed to quote something else?

If you still don't believe me just look at how many people complain about Hyperoptic's 1Gb/s service being contended as hell, yet their speedtests to the nearest hop max out to full speed!

Nope, it was generally addressing your comment but I tried to make it more relevant.

Hyperoptic as an example....
First hop to them is fine and gets 1Gb, then thats fine. They are delivering gigabit speeds, if they have a speedtest server and the customers test over their WAN connection and they get gigabit then brush it off as done.

Hyperoptic can't be held responsible for their customers being uneducated enough to stray off onto websites and onto servers that don't have as much throughput speed as they do so when a customer speedtests off a site professing to be "decent" and they get 300Mb (for example) as that is the max speed said site can carry, your comments imply hyperoptic are not delivering the speed they should be which is false.

 

[bledd]

I always get better results on speedtest.uk.net, much better site too.

Nice site