Time to Ditch Speedtest.net?
ardichoke
Icrontian
If this article is to be believed, Speedtest.net's rating methodology is pretty flawed and shouldn't be trusted.
https://ting.com/blog/internet-speed-is-not-a-subjective-measurement/
Perhaps it's time to start using https://dslreports.com/speedtest instead? It isn't as shiny, but at least it doesn't require flash... plus it gives more detailed information on the test results.
0
Comments
I cannot understand the WILD disparity in results between DSLReports and Speedtest.net.
On DSL reports, from work just now, I got 155mb down/136mb up. On Speedtest.net I got 20mb down/63mb up.
That's a massive difference...
That article is not really about the testing methodology, just that it isn't fair that small ISPs do not have a chance to be ranked on speedtest's rankings because they don't have enough customers to meet the min. req.
All speed tests are going to provide you different results. Depends on hops between you and the server, the equipment to and from, loads, blablabla.
DSL reports: 126/12
Speedtest: 110/10
Just FYI: Ookla also has pingtest.net which measures packet loss and the difference in the pings.
I feel like every time I go to speedtest.net, there is one more ad box on the page than last time I went there.
DSLR: 2.19/0.62
SpeedTest: 2.19/0.75
I'm so far out in the steppe, these took almost the same path I guess? Since, as far as I know, there is only one provider in this area.
I use speedtest.net at work when I'm making adjustments to get the best WiFi throughput. I've found it to be a good tool if not completely accurate I at least know if I'm heading in the right direction when I decide to turn off beam forming on an adapter or perhaps turn off mixed mode for straight AC wireless. The before and after is there and I find that its pretty repeatable once I set it somewhere and come back and run it again in a few minutes to check for consistency. The adds on the site have gotten silly though. I just got a Special K add in Spanish on four different panels.
AD.
Thanks to NoScript, Ghostery, and AdBlock I've never seen an ad there...or almost anywhere for that matter.
dslreports' testing methodology seems a bit more sound too... seems like it measures your speed to multiple servers simultaneously (I also like that it measures buffer bloat, though I'm unsure how accurate that measurement is). Speedtest.net's reliance on a single server to test your connection seems pretty iffy and unrealistic... plus I just tried back to back SpeedTest's and got wildly different numbers from two servers in the same city as me. One gave me a result of 26Mbps, the other gave me 96Mbps.
I didn't find much difference between the two here at home...
WELL NOW YOU'RE JUST SHOWING OFF
This thread has opened my eyes to DSLReports as an alternative. I'm going to start comparing in different scenarios this week.
Pretty close for work, not really any difference to write home about.
I also get pretty comparable results at work. I must say, I miss the absolutely sick amount of bandwidth I had available at LiquidWeb... but yeah, I really like the DSLReports test, especially the bufferbloat analysis, since that is typically the biggest "speed" problem that people have these days. No surprise, my home Comcast connection consistently gets a big fat F for bufferbloat.
I received a letter from Comcast a while back saying they've upgraded my speeds for free*, and that I only need to reset my modem. I finally got around to it today. Here's the before and after.
Dat bufferbloat. The reason I didn't bother resetting until now is that the speed is plenty for what I can do at home (if I was allowed to run a server, then I'd have a need for more bandwidth). I'd just like to pay less money for this speed.
Another company is very, very slowly rolling out fiber to neighborhoods in Chambana, but a neighborhood has to hit 50% signups before they install, which is a pretty high bar to pass. My neighborhood isn't even close. I dream of the 1000/200 Internet connection I signed up for. Again, not because of the speed, but because I wouldn't have to give another dime to Comcast.
I have started using DSLReports pretty exclusively. Thanks @ardichoke for the tip. If nothing else getting away from those awful banner adds on Speedtest. I find that DSLReports adds some detail and depth to the methodology. It isn't that Speedtest is bad, it's that DSLReports is better.
I just want to say: what Garg said. Everything that he said. Fuck Comcast, I also eagerly await another ISP nearby to roll out anything that isn't DSL.
Our new supercomputer is connected up to a phat pipe. The website wasn't happy with the performance of my remote X session, so this could be better if I tried again with elinks or got VNC installed.
Plus fucking 1. Lightspeed Internet is (very) slowly rolling out FTTH in East Lansing. Their status map has my new neighborhood marked as "Under consideration"... but the area marked stops just short of my house (we're talking about a half dozen houses short of my new house). If they roll it out in the area I will do anything in my power to convince them to run it the extra yards to get it to my house. Even if I have to pay them extra. It would be worth it just to finally tell Comcast to shove their cable up their corporate rectum.