Speedtest reading 0.5 Mbits under other tests.

[suntreader]

I have consistently found that your speed test reads approx 0.5 Mbits under www.speedtest .net or speedtest .magnet.ie ???

How do you calibrate your tests?

[Sean]

From my experience of using speedtest.net, they seem to give the peak speed rather than the average speed measured during the test, which is useful to see what the fastest your line can possibly do. However, I have even seen a few cases where speedtest.net has provided results higher than the DSL line is physically connected at, e.g. 3.5Mbps on a 3Mb DSL line where here would show 2.6Mbps, which makes me wonder how it actually calculates the speed.

The speed test here gives an average speed over the duration of the test, which will be closer to the actual speed you'll get on a download from a fast server. To see how much your speed varies during the test, go to the "Graph" tab, which appears a few seconds after the test here completes.

Another way to measure your average speed would be to download a test file from the following link (e.g. the 15MB file) and when it reaches around 90%, take note of the transfer rate shown and multiple this by 8 to get your downlink in kbps:
BT Openworld

[suntreader]

Hi Sean

I see what your saying but all the other tests agree!!!

The large file download give a reading of half what you give so I guess thats server dependent.

Speedtest.net will vary depending on which server you go for (Dub, Lim, Gal) and the speed does go up and down during the test. The final figure is not usually the highest figure reached during the test.

I'm not saying your test is any better or worse, it's just that a lot of people seem to take these tests as gospel and I wonder how accurate any of them are?

[oneill.dav]

My experience with this test has always been positive. Speedtest.net is not as accurate, in my opinion. I have always found this test to tell the truth more so than any test. Remember this: always use a domestic server. Using anything foreign, even as close as the UK, can potentially provide inaccurate results. This is due to latency.

[suntreader]

as I said I only used domestic servers (Dublin, Limerick, Galway).

On what basis do you say that this speed test is better than others. No disrespect but an opinion is something very subjective. What are the technical reasons to back up that opinion?

I should say that I come from a technical background and that is why I like to get down to the nitty gritty.

If there is no difference technically and one server in five consistently gives a different result then I question that result! I'm not saying that this test is not the most accurate I just need the technical backup to justify it.

[blacknight]

Last time I checked speedtest.net didn't publish the details of server locations and peering arrangements, whereas we've always been very open about our connectivity.

[foggy_lad]

Speed test statistics
---------------------
Download speed: 123568 bps
Upload speed: 45472 bps
Quality of service: 95 %
Download test type: socket
Upload test type: socket
Maximum download pause: 385 ms
Average download pause: 252 ms
Minimum round trip time to server: 214 ms
Average round trip time to server: 551 ms

VoIP test statistics
--------------------
Jitter: you --> server: 91.2 ms
Jitter: server --> you: off
Packet loss: you --> server: 0.0 %
Packet loss: server --> you: off
Packet discards: 0.0 %
Packets out of order: 0.0 %
Number of supported VoIP lines: 0
Estimated MOS score: 2.1



two different results within a minute of each other and a would have to say the speedtest.net test is fine for the odd test but it can vary with such wild results like 50mb or more it is not as reliable as the test here.