Speedtest.net Produces Different Results Using Different Browers

[bubba]

It's true that IE is dog slow with Java...as it is with graphics rendering. Maybe this has something to do with some of the results users are seeing with speed tests. Do a google and you will find loads of references. Here is one:

http://lifehacker.com/5286869/lifehacker-s...rome-2-and-more

And it's not just download speed that affects your faster browsing experience. It's also how fast the browser renders the page.

Apart from being slow, IE is also plagued with security problems and it doesn't comply with many web standards. Read about the Acid3 web standards check here:

http://en.wikipedia.org/wiki/Acid3

Check this to see how your browser does:

http://acid3.acidtests.org/

Both IE 8.0 and the preview release of IE 9 produce the worst web standards failure percentage of any browser.

Why do informed users still use IE?

Edited February 20, 2010 by bubba

[Beggar]

Try this.

Go into to registry to this key

HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services\AFD\Parameters

Select START -> RUN -> type regedit and then look for this key in

HKEY_LOCAL_MACHINE

then SYSTEM

then ControlSet001

then Services

then AFD

then Parameters

and there should be the following entries (these data are from WinXP SP3):

DefaultReceiveWindow and DefaultSendWindow. I have them at 76800 what is very high already - right click on the entry and enter it as DECIMAL value. Then there should be no difference between Firefox and Chrome anymore. BUT before you change the values write down the values you have now - just that you can set the values back.

As always the warning: changing the wrong things in the registry can render your system useless. So if you cannot handle it don't try it!!!

I am not sure anymore if you have to restart only Firefox or the whole system that the new values will work. Just find out for yourself. If the speed gets worse then this value 76800 was too high for your connection and you have to play around with lower values.

Actually the value depends on your connection quality to a server. So you have to increase it if my value does not work from the value you have now until it is fast enough - increase if you have a low value now (for instance 1024), decrease if you have a high level now. If the buffer value is too big then imagine a delivery service - you have one bad box in the delivery and you have to send back all of them to get one new good box. Would be nice to just send back the bad one. But this does not work. Only if you have the buffer so small that only one box fits in. But then you are/the system is/the internet is busy with all the endless deliveries for the single boxes - there is a lot of paperwork and other things to do to move them around So the better your connection and the less errors (bad boxes) the bigger the buffer can be. You should take a multiple of 1024 as value and increase in small steps.

The values for receive and send can/should be different. Sending and receiving often need different buffer sizes. So you have to do the test for up AND download -> send AND receive buffer.

[Totster]

Thanks for the post..

However, I shouldn't have to mess around in the registry to make FF faster

Totster

[higgy88]

Totster,

I'm curious. Along with the substantial differences in speed tests, do you also see differences in general surfing and opening web pages? And do you see differences in say, watching streaming videos or downloading or other bandwidth intensive activities?

[Totster]

Totster,

I'm curious. Along with the substantial differences in speed tests, do you also see differences in general surfing and opening web pages? And do you see differences in say, watching streaming videos or downloading or other bandwidth intensive activities?

You tube streams without stopping when using Chrome.. when using firefox on the same video, it stops and starts, constantly having to buffer

I did a test with a picture download from this forum, and Chrome was again faster

totster

[higgy88]

Thanks. I guess for whatever reason, your computer likes chrome more than firefox. I certainly don't know why. But if chrome works better than firefox, that would be the one I would use. I sure wish I could watch youtube videos without constant buffering, on any browser.

[welo]

So the big question for me is if there is a significant speed drop in Firefox that affects daily 'work'.

Taking totster's statement on Youtube performance into account, it seems yes.

Do you guys mind taking another speedtest website - just to rule out that the differences are due to the code in speedtest.net's benchmark.

http://speedtest.adslthailand.com/

welo

[bubba]

So the big question for me is if there is a significant speed drop in Firefox that affects daily 'work'.

Taking totster's statement on Youtube performance into account, it seems yes.

Do you guys mind taking another speedtest website - just to rule out that the differences are due to the code in speedtest.net's benchmark.

http://speedtest.adslthailand.com/

welo

I'm using a Mac with 1028 service here in Thailand. I just tried the above with the Thai server and got essentially identical results with Safari, Chrome, and Firefox (thankfully, IE is not an option on Macs). In every case, download was 850 ±3 Mbps and upload was 415 ±3 Mbps.

[ronz28]

I confirm Google Chrome is faster but it seems to surge to get the top speed and didn't always get the top speed. I hope these links will display

Chrome Chrome

http://www.speedtest.net/result/724878154.png

Firefox

http://www.speedtest.net/result/724880174.png

Opera

http://www.speedtest.net/result/724881493.png

Internet Explorer

http://www.speedtest.net/result/724882830.png

Rerun1 Google Chrome

http://www.speedtest.net/result/724883891.png

Rerun 2 Google Chrome

http://www.speedtest.net/result/724884466.png

Clear Chrome Browser Data and Rerun 3 Google Chrome

http://www.speedtest.net/result/724886132.png

Edited February 22, 2010 by ronz28

[Totster]

So the big question for me is if there is a significant speed drop in Firefox that affects daily 'work'.

Taking totster's statement on Youtube performance into account, it seems yes.

Do you guys mind taking another speedtest website - just to rule out that the differences are due to the code in speedtest.net's benchmark.

http://speedtest.adslthailand.com/

welo

Hi, test results as follows

Thai Server

FF

Chrome

HONG KONG Server

FF

Chrome

Although, I think it's important to note that ADSL Thailand uses the same software as speedtest.net, as provided by Ookla

Totster

[Beggar]

Here are my results with Firefox and Chrome - they are pretty much the same. Just change the values in the registry I mentioned before. Firefox takes the values from Microsoft, Chrome does not care about them.

Edited February 22, 2010 by Beggar

[lomatopo]

I use the NECTEC speedtest facility. Speeds reported for Firefox are 10% greater than those for Chrome. But I do like the Chrome interface and it does start up much, much faster than FF.

[Totster]

I use the NECTEC speedtest facility. Speeds reported for Firefox are 10% greater than those for Chrome. But I do like the Chrome interface and it does start up much, much faster than FF.

Ok, just carried out the test - results below

Download Speed : 1.5 Mbps

Typical Dialup 28 kbps

Best-case Dialup 56 kbps

Standard DSL 256 kbps

Your Speed 1.5 Mbps

T1 1.54 Mbps

Cable 3 Mbps

Ethernet 10 Mbps

Upload Speed: 636.08 kbps

Typical Dialup 28 kbps

Best-case Dialup 56 kbps

Standard DSL 256 kbps

Your Speed 636.08 kbps

T1 1.54 Mbps

Cable 3 Mbps

Ethernet 10 Mbps

Data

»

Date & Time = February 22, 2010 16:38:18

Download Speed : 621.7 kbps

Typical Dialup 28 kbps

Best-case Dialup 56 kbps

Standard DSL 256 kbps

Your Speed 621.7 kbps

T1 1.54 Mbps

Cable 3 Mbps

Ethernet 10 Mbps

Upload Speed: 780.78 kbps

Typical Dialup 28 kbps

Best-case Dialup 56 kbps

Standard DSL 256 kbps

Your Speed 780.78 kbps

T1 1.54 Mbps

Cable 3 Mbps

Ethernet 10 Mbps

Data

»

Date & Time = February 22, 2010 16:37:36

Totster

[bubba]

Totster - were you using Windows or Mac?

This all seems to be related to Windows OS somehow. I have now tried this using three browsers (Safari, Firefox, and Chrome) on two different Mac OSX machines, and there is absolutely no difference between the browsers (as noted in my previous posting above) on the same machine. The suggestion here about doing something with the Windows registry helped speeds suggests to me that it's not just browsers that are causing these speed differentials, it's Windows.

Any other Mac users care to try the speed tests and share your results here?

[Totster]

I'm using windoze

totster

[Crushdepth]

I've found Chrome to be (on average) marginally faster than Firefox on the Thaivisa speed test - we are talking 2.6 to 2.8 MB for Chrome vs 2.5 MB for Firefox, so not a big difference. But the Chrome results also seem more variable than Firefox. And the download speed needle really moves up in sudden jerks with Chrome.

My impression is that Chrome is just opening more concurrent connections than Firefox. Maybe it is also better at figuring out what the optimum number is. You can probably tweak that if you want.

[welo]

My PC (Windows 7 64bit, Core Duo 2 2.1Ghz) shows about 5% better results for Chrome (4.0) compared to both Firefox 3.5.7 and 3.6 (both 32bit) on speedtest.net.

I always thought speedtest.net opens only 1 connection to measure bandwidth. Now that Crushdepth brings up the issue I am questioning my own assumption. Why I think only 1? Because using torrents or download managers I usually can max out my 2MBit connection internationally, but with speedtest.net not. Still, speedtest might just use viewer connections, not necessarily only 1.

Totster, since your IE performs similarly badly as FF I'd rather rule out Firefox issues.

I also doubt it's a Javascript issue - if so, I'd consider that a flaw in speedtest.net's code which shouldn't reflect differences in JS performance.

Beggar's explanation seems very reasonable. Now that you made everybody dizzy with your speed results you HAVE TO try changing those settings

welo