I was using speedtest.net to check my speeds and was very unhappy (5mbps down and 4.5 up). I was told to use speedtest.comcast.net and had the same results.
After we reset the modem, my speeds took off. I was getting between 20-30mbps down and 5 up. Yesterday, though, my speeds went back down to 5up and 5down according to speedtest.net. For whatever reason, though, speedtest.comcast.net is still saying at least 20down and 5 up.
Why would speedtest.net and speedtest.comcast.net give me different results?
It is considered a gold standard by savvy techs for accuracy and reliability in comparison to any of those run of the mill Flash/Java based type of testing sites. It has the ability to differentiate between the "Powerboost" burst speeds and the actual sustained speeds.
The speedtest.net site has routinely not made the best choice in which server to use when I run a speedtest through them. One of the servers I see on the map of my area is hosted by comcast. When I pick that server I get similar results to the comcast speedtest. So, hover over the different test sites and look for one that is "Hosted by: Comcast".
So, hover over the different test sites and look for one that is "Hosted by: Comcast".
These will generally show better results because they are hosted right on their network's "edge" so that the packets do not have to traverse any "third party" backbone transit provider's systems. IMNSHO, the Shaperprobe tool is still more accurate and reliable as are multiple simultaneous (say four or five) "real life" large file transfers from known to be fast and closely located servers via a download manager app while observing the transfer rate. This will "MAX OUT the pipe and better reveal its true capacity capabilities.