19 replies to this topic

[Heeter]

Hi all,

Wondering what is the difference between speed test sites. I test with testmynet, then speedtest.net at the results are not even in the same ballpark. Even though speedtest.net selects a speed test server in my city, This is still the internet, and the results should be a lot more comparable.

Speedtest.net speed 24Mbps download/ Testmy.net speed 2.4Mbps download.


Thanks all,


Heeter

[CA3LE]

The test method here is completely different.  Some factors that may slow down your performance when testing here (that aren't always detected by other speed tests) include... Incorrect MTU and/or Rwin settings and slow hard drive performance.

Another factor that effects all speedtests is the route to the server.  If there is congestion along the route then you'll see slower speeds than you might expect.  We run our dedicated servers through one of the most powerful and well peered providers in the country... so if there is congestion the clients host is usually to blame (Shaw in this example).

Speedtest.net's servers are usually so close to you that route is rarely a factor... but I personally believe that people are only lying to themselves when they test off such close servers.  It defeats the purpose of this type of test.... do you want to know what your speed is to your neighbors house or do you want to know what your speed is to a normal, more realistic location.

If you have a minute read this.  Pay attention to the final paragraph.

You're only as fast as the weakest link between you and the test server.  I have done everything on my end to ensure the best possible testing environment now it's really up to the ISP's to get their customers routed through better providers and bigger, more adequate pipelines along the way.

Speedtest.net isn't wrong by the way... I just think that they test people under unrealistic circumstances.  

If you're host is doing things right you should be able to max your line out here from anywhere in North America...

:::.. CA3LE's Combined Test Results ..:::
Download Connection is:: 53574 Kbps about 53.57 Mbps (tested with 50 MB)
Download Speed is:: 6697 kB/s or 6.7 MB/s
Upload Connection is:: 4159 Kbps about 4.2 Mbps (tested with 2.9 MB)
Upload Speed is:: 520 kB/s
Tested From:: http://testmy.net (Dallas, TX USA)
Test Time:: 2011-02-13 12:39:33 GMT -7
Validation:: http://testmy.net/db...N2IVEL&dual=yes
User Agent:: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US; rv:1.9.2.13) Gecko/20101203 Firefox/3.6.13 [!]

I'm like 1400 miles from the server by the way.

[tdawnaz]

yeh just tested at  speedtest and they have me at 25 dn and 3 up...bahahahaha  i wish!!...but no...not even close...i pay for 12/2...and what i get here is...



which is great for me...i prefer to know the truth...and the truth is...i'm not gonna get 25 dn without paying for it

[CA3LE]

tdawnaz, on 13 February 2011 - 03:42 PM, said:

yeh just tested at  speedtest and they have me at 25 dn and 3 up...bahahahaha  i wish!!...but no...not even close...i pay for 12/2...and what i get here is...



which is great for me...i prefer to know the truth...and the truth is...i'm not gonna get 25 dn without paying for it

Definitely a great example.  Thanks tdawn

[Heeter]

So, my ISP is selling me a package up to 15 Mbps (Shaw Cable - Edmonton, Alberta, Canada), it is not even close with testmy.net, should I be concerned. Should I be raising a stink?

A long time ago, I pinned them on the fact that they optimize their results on a java script for windows. I use Linux which showed completely different results. At one point, I tested one machine with 2 different hard drives ( one with Vista, other with Debian). I would test with the windows Vista HD in the machine, I would get the speeds that my ISP promised. Then I would boot from the debian Hard Drive from the same machine, and it would be about half. The ISP tech admitted that the javascript on their own speedtester site was optimized for windows machines.

Thanks for detailed response, CA3LE, TDAWNAZ


Heeter

[TriRan]

i wouldn't be... if i were downloading from a server in canada it would probably yield similar results

[xs1]

Yea speedtest.net makes you feel pretty good , but testmy.net is far more accurate. Ill give you an example on my  crap 15 / 2 advertised line ;  As you can see their test gives me FAR more than advertised speeds.

Now there are days it runs well and i see 20 ; but 25 , 26 is really pushing it. I've seen as high as 28 from speedtest.net...  on my 15 meg line lol

Testmy.net :

Download Connection is:: 13295 Kbps about 13.3 Mbps (tested with 25 MB)
Download Speed is:: 1662 kB/s or 1.7 MB/s

Upload Connection is:: 1359 Kbps about 1.4 Mbps (tested with 2.9 MB)
Upload Speed is:: 170 kB/s



Speedtest.net:

Download : 24.28 Mb/s
Upload : 1.65 Mb/s

seen here

[CA3LE]

Heeter, on 13 February 2011 - 09:00 PM, said:

So, my ISP is selling me a package up to 15 Mbps (Shaw Cable - Edmonton, Alberta, Canada), it is not even close with testmy.net, should I be concerned. Should I be raising a stink?

Well, are you experiencing slow downloads to most servers?  If yes, I would raise a stink.  Maybe run a scheduled test and see what times you're slowing down.  These problems are usually associated with peak traffic hours.  After you've tested you can give your stats link to Shaw.

Heeter, on 13 February 2011 - 09:00 PM, said:

A long time ago, I pinned them on the fact that they optimize their results on a java script for windows. I use Linux which showed completely different results. At one point, I tested one machine with 2 different hard drives ( one with Vista, other with Debian). I would test with the windows Vista HD in the machine, I would get the speeds that my ISP promised. Then I would boot from the debian Hard Drive from the same machine, and it would be about half. The ISP tech admitted that the javascript on their own speedtester site was optimized for windows machines.

We don't have that problem here.  Firstly, I'm a linux guy myself.  Secondly, I program with compatibility on the forefront of my mind.  I try and make sure that my users don't require anything special to run the tests here.  Any OS with the vanilla browser will work properly here.  Another thing that this site does is show the difference between browsers (other speed tests don't seem to notice).  Third... I extensively test my programs in literally EVERY browser.  There are a couple of known browser issues... the PS3 browser (piece of crap) simply doesn't work and Opera (another PoS) doesn't perform well, I personally can only pull 20Mbps out of it on my 50Mbps line (Mac Version).   The methods that are employed here enable this test to benchmark browsers as well as your internet.  Short comings of browsers like opera become very apparent when you test here.    Over the years I've seen the balance of power shift from IE to Firefox to Chrome back to Firefox... etc.  Depends on the build really.  If you run tests back to back using different browsers you'll see what I mean.

Spoiler

Little side note: The tests here are really unique to the internet.  I took a completely different path/approach to my internet testing solution. It's going to take thousands more hours of work for me to enable the system to tell the client the exact issue that's slowing them down (instead of just showing you a number and have you troubleshoot)... but that's my end goal with this project.  I've been working on this for over ten years and I still don't see any end in sight.  The internet changes too quickly for it to ever be finished.

Heeter, on 13 February 2011 - 09:00 PM, said:

Thanks for detailed response, CA3LE, TDAWNAZ


Heeter

No problem, that's what we're here for.  Please tell your friends about the site.  This is 100% word of mouth.

- Damon

xs1, on 14 February 2011 - 04:19 AM, said:

Yea speedtest.net makes you feel pretty good , but testmy.net is far more accurate. Ill give you an example on my  crap 15 / 2 advertised line ;  As you can see their test gives me FAR more than advertised speeds.

Now there are days it runs well and i see 20 ; but 25 , 26 is really pushing it. I've seen as high as 28 from speedtest.net...  on my 15 meg line lol

... Condensed ...

Another good example.  Thanks Nick

-D

[xs1]

CA3LE, on 14 February 2011 - 04:24 AM, said:

Well, are you experiencing slow downloads to most servers?  If yes, I would raise a stink.  Maybe run a scheduled test and see what times you're slowing down.  These problems are usually associated with peak traffic hours.  After you've tested you can give your stats link to Shaw.

Only problem with this idea is the same my ISP did to me. They'll claim that the server is too far from himself to guarantee speeds. I believe their exact phrase was " We only guaranty speeds throughout the local networks ; anything over the boarders are not guaranteed "   Bottom line is to protect their best interest they will tell him to test locally.

[CA3LE]

xs1, on 14 February 2011 - 04:29 AM, said:

Only problem with this idea is the same my ISP did to me. They'll claim that the server is too far from himself to guarantee speeds. I believe their exact phrase was " We only guaranty speeds throughout the local networks ; anything over the boarders are not guaranteed "   Bottom line is to protect their best interest they will tell him to test locally.

Yeah, companies always get away with stuff in the fine print.  The best we can really do is try to educate people on their practices and recommend ISPs with better practices.  I just hope that the information here has and will continue to help make the internet a better (faster) place.  

[TriRan]

CA3LE, on 14 February 2011 - 04:37 AM, said:

Yeah, companies always get away with stuff in the fine print.  The best we can really do is try to educate people on their practices and recommend ISPs with better practices.  I just hope that the information here has and will continue to help make the internet a better (faster) place.  

no such thing as too fast

[CA3LE]

TriRan, on 14 February 2011 - 08:34 PM, said:

no such thing as too fast

Werd!

[haifa7]

Nice................

[tdawnaz]

CA3LE, on 14 February 2011 - 04:37 AM, said:

Yeah, companies always get away with stuff in the fine print.  The best we can really do is try to educate people on their practices and recommend ISPs with better practices.  I just hope that the information here has and will continue to help make the internet a better (faster) place.  

Attached Thumbnails

• 

[TJS]

Heeter, on 13 February 2011 - 10:52 AM, said:

Hi all,

Wondering what is the difference between speed test sites. I test with testmynet, then speedtest.net at the results are not even in the same ballpark. Even though speedtest.net selects a speed test server in my city, This is still the internet, and the results should be a lot more comparable.

Speedtest.net speed 24Mbps download/ Testmy.net speed 2.4Mbps download.


Thanks all,


Heeter

Just thoguht id continue on this












Strange how south africa is not much slower then i get to parts of the us, when its half as far again the distance from dallas tx?

And newzealand at over 2 times as far

I feel somewhat skeptical that its my isp to blame for connections to the mid/south of america being slow when newyork hits faster. it seems as though the poor routing is mid/south America to Europe?  I could be wrong though  

Please take note of the distances on the test results

[tdawnaz]

TJS...where are your results for tmn...not sure why you are posting speedtest results here

[CA3LE]

TJS, on 06 May 2011 - 04:27 PM, said:

Just thoguht id continue on this
...

Please read TestMy.net scores ONLY

[VeryKen]

Here's something ridiculous I found while browsing Comcast's offers. It's a test site that actually gives you a MUCH LOWER score, but it also has a link redirecting you to another test site of theirs assuming that existing customers don't see any difference. I've been a Comcast customer subscriber for many years. It's a love-hate relationship.

http://business.comc....aspx#speedTest
My results:
4.23Mbps Download
2.45 Mbps Upload

Compare to their "existing subscriber" test site:
28.80 Mbps Download
4.11 Mbps Upload

Which is very similar to SpeedTest... basically around 30Mbps/ 4Mbps

My numbers here (testmy) are:
4.38 Mbps Download
3.30 Mbps Upload

All performed just now, within minutes apart of each other.

[xs1]

VeryKen, on 17 May 2011 - 01:16 PM, said:

Here's something ridiculous I found while browsing Comcast's offers. It's a test site that actually gives you a MUCH LOWER score, but it also has a link redirecting you to another test site of theirs assuming that existing customers don't see any difference. I've been a Comcast customer subscriber for many years. It's a love-hate relationship.

http://business.comc....aspx#speedTest
My results:
4.23Mbps Download
2.45 Mbps Upload

Compare to their "existing subscriber" test site:
28.80 Mbps Download
4.11 Mbps Upload

Which is very similar to SpeedTest... basically around 30Mbps/ 4Mbps

My numbers here (testmy) are:
4.38 Mbps Download
3.30 Mbps Upload

All performed just now, within minutes apart of each other.

Must be your routing. As far as "speeds" go the site is decent. Upload is dog slow tho.. Best i could pull out of it was 16..


First Screenshot :   Comcast Business

Second Screenshot: Testmy.net

Attached Thumbnails

• 
• 

[thomaswilliam222]

However you try to test the speed of your site in different speed test sites you will not be able to get the accurate speed. So its better to test your site with some site speed testing softwares.