Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[jeffbb]

Hi   
     I have a query about speed testing .
With a synch speed of 6.8Mbps I have had the following results from 4 different speed testers
Speedtest.net   4436Kbps
Max                   2254Kbps
Bt                      4175 Kbps
Thinkbroadband 2137 Kbps.
Speedtest          4305 Kbps     To complete check over the 5 or so minutes to do this .
During these test I was using both Drspeedtouch  view the data . on all occasions the graph reflected the reported speeds .
I have done MANY speed checks using these sights .   Max and thinkbroadband have never gone above 2.3 Mbps even though Speedtest.net has shown up to 4.7Mbps.
When my speed drops to about 400Kbps  or less on speedtest.net then both Max and thinkbroadband show a similar result s.
Below 300Kbps Bt speed test tells me "This test was not conclusive and further testing is required." I have not sucessfully done the 2nd test yet

At about 2Mbps on speedtest.net BT speed test is in agreement but  Max and thinkbroadband show about 60 to 70% of the speed . again this is reflected on Drspeedtouch graphs . The results are also confirmed using Task Manager .
Confused .Jeff 
 

[roseway]

I don't really think that you can conclude anything from differences between different speed test sites. The results will always depend on things which are out of your control, such as server loading, network congestion and ISP traffic management. And sometimes there's something about a particular site which makes it always deliver poorer results for a particular person. In addition, the results will always be limited by your current IP profile, and it appears from your figures that your IP profile was around 4500 when you did the tests.

[kitz]

>> Thinkbroadband 2137 Kbps.

The TBB speedtest uses non-standard http ports, so its possible that traffic to the TBB speedtester is being more heavily shaped by tiscali than a speed tester which uses port 80. 
I've seen mention of this before with Tiscali and its one of the reasons why TBB offers the selection of a port 80 tickbox option.

[jeffbb]

Hi
tried port 80 . Its made no difference . both broadbandmax and thinkbroadband give significantly lower results at higher speeds . As BT and speedtest agree I will use those as a reference .Dr speedtouch performance graph reflects the reported speeds in all cases .
Regards Jeff

[kitz]

Sorry dont know then :/

May be an idea to grab something like Netmeter and see how "real life" downloads perform.

[jeffbb]

Hi
 Tried Netmeter  Its a nice piece of software . It tends to confirm the speed readings from each of the sites . speedtest and BT give a higher download speed readings and the  netmeter graph reflects that .BBmax and thinkBB show lower download speed  readings they  are also reflected in netmeter. The only thing to be aware of  is that when the download speeds   are Low then netmeter reports the highest reading during the test ,but you can still interpolate the mean from the graph.
Regards Jeff

[kitz]

>> but you can still interpolate the mean from the graph.

Yep I look at the spikes (if any) whilst its downloading.  If certain sites give better results than others then in this instance it could be that certain links out on to the internet are more congested than others.

Trying to think of an easy way of putting it - but from the ISP they can have many different links out on to the internet depending on where the traffic is being routed to.  They will normally install (several) GB links out to different transit providers.  It is where congestion can take place and I know of an ISP atm which is waiting to install another Gigabit link to one of their transits.  To show what I mean look at this diagram below of Plusnets routing which shows they have several different peers/transits - depending on where the traffic is going.

The other alternative is that they are giving certain sites priority over others.  Its not unknown for some ISPs to shape all traffic, but let traffic to certain speedtest sites pass through unhindered.   Bit surprised that TBB wouldnt be on that list though. 

[jeffbb]

Hi
   Here is something odd . I am in process of setting up anew PC  with Vista Operating system !. more of that in another thread . On the new machine ALL my different speed test all match more or less? that is BT, speedtest, thinkBB and BBmax  all agree on Download speeds . 
regards Jeff

[kitz]

Sometimes there is a little difference between speed tests on various Operating systems.. possibly due to the way it handles MTU, but it should more or less be the same like you say.

[jid]

Hi
   Here is something odd . I am in process of setting up anew PC  with Vista Operating system !. more of that in another thread . On the new machine ALL my different speed test all match more or less? that is BT, speedtest, thinkBB and BBmax  all agree on Download speeds . 
regards Jeff

I found that with Vista too. You find Vista (may) give different ping times to XP machines at certain times too.

http://www.kitz.co.uk/adsl/vistaMTU.htm

Have a look at the above link and make sure MTU is set to 1458, which is Tiscali's recommended MTU setting.

Regards

Jamie

*edit* Kitz was typing the same time as me

[orainsear]

I'm with Zen and when I perform a speed test I tend to use Zen's locally hosted files (10MB and 100MB on their servers).

Like you I have seen differences between web based speed testers, however I find that speedtest.net seems to consistently provide a reasonably good reading that correlates with the Zen test speed.

With speedtest.net the throughput speed is accurate, however the ping times reported will from time to time seem a bit high but I think that might be due to busy servers at their end.

[jeffbb]

Hi
I have not yet finally set up my new machine ,I will be doing that asap .
JID the recommended MTU setting of 1458 I am confused . I have followed KITZ method of checking on my present machine .This  has XP operating system using the method as described I get good results with a setting of 1500. No obvious improvements by dropping to 1458.
When I finally set up my new machine I will TRY to follow the instructions on the link you have given me .
Thanks Jeff

[kitz]

If you have checked your packet fragmentation size and all is well at 1500, then best leaving it at that.
I think there must be variance across tiscalis network as it can depend on how theyve set up their own dslam if their on LLU.

[jeffbb]

Thanks for the reply I will stick with it at 1500
Jeff