is Testmy.net reliable?

[shariq607]

on all other speed test sites i average around 25-35 mb to west coast US servers but slightly lower to east coast US servers but on testmy.net to San Jose server i only average 10-20 Mb or only half of what speedtest.net shows but i find it odd because i can confirm with real world single threaded downloading from servers in states that i get much higher than what testmy.net shows. also performing speed tests on various streaming services from the states shows speeds higher than testmy.net which leads me to believe either the servers of testmy.net are choked or they do not have enough capacity.

Edited November 22, 2013 by shariq607

[shariq607]

:::.. Download Speed Test Result Details ..:::

Download Connection Speed:: 13037 Kbps or 13 Mbps

Download Speed Test Size:: 25 MB or 25600 kB or 26214400 bytes

Download Binary File Transfer Speed:: 1630 kB/s or 1.6 MB/s

Tested At:: http://TestMy.net Version 13

Validation:: http://testmy.net/db/1SECrLc

TiP Measurement Summary:: Min 6.44 Mbps | Middle Avg 14.49 Mbps | Max 15.12 Mbps | 7% Variance

TiP Data Points:: 6.76 Mbps, 14.41 Mbps, 14.76 Mbps, 13.91 Mbps, 13.59 Mbps, 14.06 Mbps, 14.41 Mbps, 14.23 Mbps, 14.76 Mbps, 14.93 Mbps, 14.93 Mbps, 15.12 Mbps, 14.76 Mbps, 14.59 Mbps, 14.4 Mbps, 14.57 Mbps, 14.94 Mbps, 13.91 Mbps, 6.44 Mbps

More Stats:: http://testmy.net/compID/1015826012419

Test Time:: 2013-11-22 14:41:49 Local Time

Location:: Bangkok, TH TH >> Destination:: San Jose, CA US

1MB Download in 0.63 Seconds - 1GB Download in ~11 Minutes - 233X faster than 56K

This test of exactly 25600 kB took 16.099 seconds to complete

Running at 385% of hosts average (CAT Telecom public company Ltd http://testmy.net/hoststats/cat_telecom_public_c)

User Agent:: Mozilla/5.0 (Windows NT 6.2) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/31.0.1650.57 Safari/537.36 [!]

[keeniau96]

Here in Phuket we did quite a bit of testing of the various speed-test sites and concluded that many are prone to overly high numbers due to: 1) gamed by the ISPs, 2) cached data actually in country. Gamed you say! Some of the ping tests returned times that exceeded the speed of light when calculating time:distance. Take a look at some of our results: http://www.khunwoody.com/phuket-internet-speed/

Testmy.net is probably imperfect (they all are) but it is the most reliable that we found.

[Pib]

I expect the other speedtesters you refer to are Flash/Ookla-based speedtesters like Speedtest.net...and these type of speedtesters usually do "multi-thread" testing when they are not being fooled by local cache servers and giving faster-than-light ping times and download speeds like the test server which is in far-off farangland is just across the soi (and the local cache server may be across the soi).

Testmy.net apparently defaults to "single-thread/linear" testing but you can set it to do multi-thread testing. See this Q&A about "What Do My Results Differ from Speedtest.net / Ookla Speed Tests? Within this Q&A you see where you can set mytest.net to do multi-thread testing versus linear testing...and of course it seems the mytest.net utilizes a tougher test in testing various random data/file sizes where I think Speedtest.net/Ookla testers have less tough testing...maybe using large sequential files.

Anyway, I'm on a True DOCSIS 15Mb/1.5Mb plan here in Bangkok...costs me Bt699/mo....if I go to testmy.net home page and click the Test MyInternet button I usually pull around a 2.5Mb download speed with their default/single-thread/linear tester. Now if I set the test to multi-thread testing I pull much higher downlaod speeds....like this test I ran at 7:15pm Bangkok time to Dallas where I got a 7.5Mb download speed, but I could see this test was less tough than the classic/linear test and of course it was multi-threaded.

[shariq607]

I expect the other speedtesters you refer to are Flash/Ookla-based speedtesters like Speedtest.net...and these type of speedtesters usually do "multi-thread" testing when they are not being fooled by local cache servers and giving faster-than-light ping times and download speeds like the test server which is in far-off farangland is just across the soi (and the local cache server may be across the soi).

Testmy.net apparently defaults to "single-thread/linear" testing but you can set it to do multi-thread testing. See this Q&A about "What Do My Results Differ from Speedtest.net / Ookla Speed Tests? Within this Q&A you see where you can set mytest.net to do multi-thread testing versus linear testing...and of course it seems the mytest.net utilizes a tougher test in testing various random data/file sizes where I think Speedtest.net/Ookla testers have less tough testing...maybe using large sequential files.

Anyway, I'm on a True DOCSIS 15Mb/1.5Mb plan here in Bangkok...costs me Bt699/mo....if I go to testmy.net home page and click the Test MyInternet button I usually pull around a 2.5Mb download speed with their default/single-thread/linear tester. Now if I set the test to multi-thread testing I pull much higher downlaod speeds....like this test I ran at 7:15pm Bangkok time to Dallas where I got a 7.5Mb download speed, but I could see this test was less tough than the classic/linear test and of course it was multi-threaded.

testmyCapture.JPG

I get around 10 to 20 mb on test my.net server to San Jose using single thread but significantly higher using multithread.

[Pib]

That's good...goes to show again how various speedtesters can be different in how they test...some do heavy duty testing, some light duty, some single thread, some multi thread, some are easily fooled, etc.

Sent from my Samsung S4 (GT-I9500)

[Pib]

This Link at the Speedtest.net website explains how they test....it says they use up to four threads to run a speed test. Of course multi-thread speed testing can make you feel good with your download results assuming the server/service you are connecting to supports multi-threading. Example: It's my understanding most streaming video sites do not support multi-threading...just single threaded operations.

The link also says Speedtest.net prevents caches from throwing off results by appending random strings to each download...maybe this is a change to Speedtest.net...but I'm not sure that is preventing caching in many cases. But I have noticed from using Speedtest.net yesterday and today that I'm not getting faster-than-speed of light ping times (like 10ms to the U.S. west coast where it should be around 200-250ms) and "usually" the download speeds in the 3 to 6Mb ballpark is very similar with what I'm getting when using testmy.net "multi-thread" testing to the U.S. west coast. However, I just did one more Speedtest.net test to San Francisco and got a 16.5Mb download speed with a valid ping time of 222ms...but I'm not believing that international download speed.

Take speed test results with a BIG grain of salt, especially if it's a flash/Ookla-based tester, not to imply all such testers are easily fooled...guess it all depends on their programming to minimize skewed/false results.

[Duvidl]

This Link at the Speedtest.net website explains how they test....it says they use up to four threads to run a speed test. Of course multi-thread speed testing can make you feel good with your download results assuming the server/service you are connecting to supports multi-threading. Example: It's my understanding most streaming video sites do not support multi-threading...just single threaded operations.

The link also says Speedtest.net prevents caches from throwing off results by appending random strings to each download...maybe this is a change to Speedtest.net...but I'm not sure that is preventing caching in many cases. But I have noticed from using Speedtest.net yesterday and today that I'm not getting faster-than-speed of light ping times (like 10ms to the U.S. west coast where it should be around 200-250ms) and "usually" the download speeds in the 3 to 6Mb ballpark is very similar with what I'm getting when using testmy.net "multi-thread" testing to the U.S. west coast. However, I just did one more Speedtest.net test to San Francisco and got a 16.5Mb download speed with a valid ping time of 222ms...but I'm not believing that international download speed.

Take speed test results with a BIG grain of salt, especially if it's a flash/Ookla-based tester, not to imply all such testers are easily fooled...guess it all depends on their programming to minimize skewed/false results.

What this entire thread basically tells me is that it is pretty much a waste a of time to check your internet speed with any available speedtester. You might as well just try to connect to the site you want and see how it goes.

Edited November 25, 2013 by Duvidl

[Pib]

Yeap, speedtesters should only be one factor in determining how good (or bad) your internet connection is....speedtesters don't tell the whole story but they tell an important part.

[david78888]

I really like your way of expressing the opinion and sharing the information.

[muratremix]

Because Thailand ISPs have pop nodes in West coast of USA (San jose and nearby), speedtest to that area will give better performance.

If you test against east coast, it will be routed through USA by some cheap ISPs or worse, via London link (Because Thai logic!).

For example my server in kansas city, mo, gets routed through london, which is so lame (and slow)

On the other hand, IPV6 through he.net singapore routes it correctly (west coast -> kansas) and pretty fast.

For example:

192.3.121.138 (ColoCrossing – Chicago, IL)

Goes via France

23.105.138.10 (NY) goes via san jose (correct path)

Whoever configure those bgp routes in Thailand (True, TOT, Cat and others) must be removed from their position immediately.

They are worse in managing network traffic compared to real world traffic (driving)