What's with my results?

[AlanMacI]

I'm really struggling to figure out what's going on... According to testmy.net I'm getting<1Mb download speed (running the test every 5 mins x 10 times). According to Google/Speedtest/Fast I'm averaging about 30. Why are the testmy.net figures so off? Am I doing something wrong? All tests run on same PC, same location, approx same time.

[Pgoodwin1]

Read all the tabs on this page. https://testmy.net/legit-speed-test
 

 

[AlanMacI]

Ok, I've just run the automatic test again (repeat every 5 mins x 5 times, default settings). My results were very similar to my last run - consistently <1Mb, with a spike or two distorting the average. 

 

According to general consensus the minimum required speeds for streaming are:

720 - 2.5 Mbps

1080 - 5 Mbps

2k - 10 Mbps

4k - 20Mbps

 

I've gone to the ‘Costa Rica in 4k’ test stream on YouTube (https://youtu.be/LXb3EKWsInQ?si=M4bTYIg4oLYl09n) - streams fine at 720 and 1080/60fps, stutters considerably at 2K and forget about 4K.

 

If, according to testmy.net, my true download speed is 600kbps, how can I stream 1080 smoothly?

[AlanMacI]

Run another set of tests... two computers next to each other on the same network:  run the automatic test (repeat every 5 mins x 5 times, default settings) one shows a median of ↓1.07/↑22.5 Mbps, other shows ↓14/↑16 Mbps?

 

[CA3LE]

Hi @AlanMacI

 

Sorry for the slow reply, I've been watching your topic develop since you first posted.

 

I just replied to a similar PM.

 

Quote

I have been using testmy.net speed test tool ever since I learned of it about a year ago. Previously I have used fast.com or speedtest.net. When I have compared these 2 with testmy.net results they have been similar although testmy.net has been consistency a little lower. Based on the additional information I get from testmy.net, I feel it provides a more realistic result. Lately, however, I started to see a big discrepancy between these 3. While fast.com and speedtest.net have remained consistent with each other at around 20-30 Mbps, testmy.net has been between 1-3 Mbps. I still feel testmy.net is more accurate as I have been experiencing what appears to be a slowdown which has been affecting streaming and slower response with other internet activities. I contacted my internet service provider Xplore, which is a Canadian satellite internet provider, and they had me do a speed test using speedtest.net which is their preferred tool and since it showed a speed over 30 Mbps they said the problem was not theirs. Testmy.net showed a speed of around 2 Mbps at that time. My questions, if you are able to help me with this are, why are testmy.net's results now so different than the other 2 and why are the other 2 still showing higher speeds when I am experiencing issues that appear to be speed related? I appreciate any thoughts you may have on this issue. Thanks

 

My Response:

 

This is the most common feedback I hear.  TestMy.net's results being lower than other tests.  Where the user is experiencing an obvious issue with speed but only TestMy.net seems to notice.

 

This is most likely because of how the data is being requested, sent and received.  TMN defaults to a single thread linear loading test.  Where the data is loaded in line, like reading a page from top to bottom.  Every other speed test I've seen defaults to a multithread process.  Where the data as a whole is made into chunks and then sent and received by multiple process.  So imagine reading the same page but being able to read 8 chunks at time.  So each individual chunk may transfer slower but when they're combined they make a seemingly faster speed.

 

I have a new version and the multithread process work amazingly well (visit My Settings and toggle Beta "ON", after enabled visit Tune ☆ to enable Multithread).  I've seriously considered making that the default because of how smooth it feels.  But that isn't the right thing to do.  What sets TestMy.net apart from the rest is that difference, the Linear test method obviously works best for overall performance tuning.  I don't want to simply give you the result that everyone else gives.  I want you as my visitor to understand that difference and what it means.

 

If I have a connection that scores poorly on linear, it's possible it can score really good on multithread.

 

It's also possible for a connection to score poorly on multithread but score really good on linear, though less likely.  I see this on wireless when the connection is limited.  I might see over 10 Mbps linear but only 0.25 Mbps multithread... like all the performance is sucked away when it tries to open many connections.

 

But if you score well on linear... it's usually a sign of a healthy connection.  The same can't be said about multithread because the aggregate speed may mask the fact that individual threads are running slower.

 

Run a few tests with the beta ON.  Enable multithread and run a few more tests.  Maybe also visit Mirrors and select a different test server for good measure.  Get more data points, then we can assess your situation better.

 

- Damon - TestMy.net