"That speedtest app uses ookla, which has multithreading enabled. This means it will open many connections to the target server, not just one. It also disounts the bottom 30% and top 10% results when giving you a result. When you download something, it only opens a single connection to the download server. So basically it's NEVER representative of your actual internet speed on a SINGLE connection. That is why every ISP recommends it - it makes them look good. To see the difference use testmy.net - which does not have either of those issues and gives you real time results that are actually accurate. I learned this the hard way trying to stream and debating with my ISP"

basically this comment is saying that the normal speedtest.net is not very accurate for real speeds?

i understand speedtest.net speeds are not 100% accurate, but i thought they were within a small range of error? maybe they have quite a large range of error?

based on this video



This post was edited by noob_whacker on Sep 14 2016 07:51am

Download :: 117.8 Mbps 14.7 MB/s up arrow Upload :: 8.5 Mbps 1.1 MB/s
TiP Summary - Minimum :: 54.99 Mbps | Middle :: 142.22 Mbps | Maximum :: 195.59 Mbps

vs

http://www.speedtest.net/my-result/5630648243

/edit
oh whoops missed the LTE part.
in any event it depends on teh server your connecting to in the end so no matter your speeds your not guaranteed them unless the server supports it.

This post was edited by King Atrhur on Sep 14 2016 08:13am

No idea



Easy to find out with wireshark running. Couldn't care less if it were true. Speedtests API allows for downloading large static generated images of different sizes and times the speed.



If all threads were downloading at the same time you would have less than optimal results to begin with so why would they bother to shave results. If you have 10mbits of bandwidth and download using 10 threads, you will get 10mbits total bandwidth but each thread will only achieve 1mbit per second. You can see where his comment doesn't make sense. Now this is from the point of view that each thread is doing the test on its own. If the threads or co-oping the test then it may complete faster than a single thread if the carrier is throttling the single connections or if there are routing problems.



Who the fuck downloads using the browsers internal download manager. Axel supports concurrent connections on linux, and firefox has plenty of addons such as "downthemall" which allow concurrent downloads and advanced options in their download manager.

Basically the guy is a fuck wit who probably doesn't even verify his claims. Also do not forget routing effects speed as well. I get 100mbit to servers but towards texas I can only get around 20mbit. To further prove my point, the "testmy.net" website says I have 40mbit per second internet, so if that is the case why can I download games at 12megabytes per second (100mbit). He could of very easily been testing to a good routed server (which speedtest attempts to find) and when downloading the youtube video to a bad routed server.

Now don't get me wrong, it is possible to have a total bandwidth cap of 100mbit but the carrier limits each connection to a specific lesser bandwidth, but that is also the carriers fault not speedtests.

Edit:: Watched the of the video. At like 8 minutes he basically explains routing problems. His problem shouldn't be about speedtest but rather the carrier so I don't know what the video guy is ragging on speedtest so much when it is the carrier scewing results (if they even are, it could be a bad radio for all we know).

Basically the commenter you quoted is a dipshit.

This post was edited by AbDuCt on Sep 14 2016 10:56pm

thanks abduct for the input
and atrhur

yea youtube comments are usually brainless comments, but was curious if there was any validity to his claim

This post was edited by noob_whacker on Sep 14 2016 10:53pm

so speedtest says
4mB down
when i down
its 4mB
must be wrong
it's like the moon
how it get there
can't explain that