speedtest.net Vs reality?

mordeith

This is probaly a stupid question but I was wondering what the deal is with speedtest.net. When I test my connection (2 Meg) I get close to 1755kbps (down) yet the quickest speed I've ever seen when actually downloading something is only around 250kbs. Is that because the speedtest.net result is just an optimal figure?
Thanks

bottletops

mordeith wrote: »

This is probaly a stupid question but I was wondering what the deal is with speedtest.net. When I test my connection (2 Meg) I get close to 1755kbps (down) yet the quickest speed I've ever seen when actually downloading something is only around 250kbs. Is that because the speedtest.net result is just an optimal figure?
Thanks

I suspect your mixing up Kb and KB
There are 8 Kilo bits (Kb) in a Kilo Byte (KB)

250KB * 8 = 2000Kb = 2meg broadband!!

Your downloads are shown in KB but Speedtest.net displays results in Kbs!!

ntlbell

Also, as I have stated many times for any of these "speed tests" it's a bit of a false economy, You're only testing you're speed against that one server.

The same goes for irishisptest from blacknight or 365's one you test against that one server, but at least testing against an irish hosted site especially when in INEX gives you an idea of your speeds in Ireland.

A good a test is just pulling something from one of the many irish ftp servers such as ftp.heanet.ie ftp.esat.net

There is nothing accurate or scientific about these tests

mordeith

Okay so what you're saying is that my true connection speed is closer to 1/8 of the maximum 2 Meg download speed? Is that normal? I realise that speeds are dependent on distance to the exchange, wiring etc.

mordeith

ntlbell wrote: »

A good a test is just pulling something from one of the many irish ftp servers such as ftp.heanet.ie ftp.esat.net

Thanks for the tip. I'll try that and see what results I get

ntlbell

mordeith wrote: »

Okay so what you're saying is that my true connection speed is closer to 1/8 of the maximum 2 Meg download speed? Is that normal? I realise that speeds are dependent on distance to the exchange, wiring etc.

250KB's give or take is about right for a 2mb product

macrubicon

mordeith wrote: »

Okay so what you're saying is that my true connection speed is closer to 1/8 of the maximum 2 Meg download speed? Is that normal? I realise that speeds are dependent on distance to the exchange, wiring etc.

As the guys above have been saying it's the differnce between bits and bytes....

250KB/s is 250 Kilo Bytes per second

There are 8 bits in a byte so...

250*8 = 2000kb/s = 2000 Kilo Bits per second

Big "B" = Bytes
Little "b" = Bits

Line speeds are almost always in bits and downloads in bytes. Confusing, but there ya go !

Sounds like you are getting max speed from a 2Mb link

tuxy

mordeith wrote: »

Okay so what you're saying is that my true connection speed is closer to 1/8 of the maximum 2 Meg download speed? Is that normal? I realise that speeds are dependent on distance to the exchange, wiring etc.

No, when advertising broadband they give figures in bits but most software for downloading measures in bytes.
Although I seriously doubt your isp calls the service you are on 2 Meg as there is no such thing as a "Meg". 2 Mb(megabit not megabyte) would be the correct term.

mordeith

mordeith wrote: »

This is probaly a stupid question but I was wondering what the deal is with speedtest.net. When I test my connection (2 Meg) I get close to 1755kbps (down) yet the quickest speed I've ever seen when actually downloading something is only around 250kbs. Is that because the speedtest.net result is just an optimal figure?
Thanks

Those online download meters give you a reasonably good estimate of your data throughput. ie What you will see in most programs as your indicated download speed.

They don't however give you your actual line speed or the raw data that is going through your line.

There is overhead subtracted from your actual DSL speed of 2048 kb/sec. (2 meg broadband).

Your indicated speed is 1755kbps or around 219KB/sec
1755/2048kbps * 100 = 86%

In other words, your practical maximum download speed is around 86% of your listed line speed. This is normal for DSL. Most other forms of broadband will have a similar overhead.

The missing 15 or so percent is used for routing and transporting the data payload.

long_b

Also at the risk of sounding dense, how come the speed the browser works at is never as good as the speedtests ? Using IE and Firefox ?

ntlbell

long_b wrote: »

Also at the risk of sounding dense, how come the speed the browser works at is never as good as the speedtests ? Using IE and Firefox ?

I'm not really sure what you're asking.

But if you're asking how come in general your download speeds don't match the test speeds are pretty explained in my post above.

You're only testing against that _one_ server.

macrubicon

Some of it comes down to how the server are intended to be used....

A general web server is set up to server many thousands of people what are usually small files... Text traditionally is quite small to process and you may notice that a lot of the images, rather than being one big image are made up of many small ones.

The specialist servers for speed tests are dedicated to the task usually. They transfer large files to gauge the speed and have to handle many concurrent tests.

They are also ususally placed in an area of the network wherre they are connected "directly" to the ISP's backbone making the traffic path to the server shorter for users.

In short - you should not expect the speed from the speed test servers to be your every day speed.... in fact you don't actually want that !

When you browse the speed test site you want to test out the max's of your link. You are probably doing nothing else with your PC.

When you are browsing normally, you may have many windows or tabs open, each taking it's own slice of bandwidth, each able to get what it needs when it needs it.

If they were all to use the max all the time, you would end up with windows or tabs lagging as they have to wait for another task to finish for them to get bandwidth etc.

long_b

long_b wrote: »

Also at the risk of sounding dense, how come the speed the browser works at is never as good as the speedtests ? Using IE and Firefox ?

Speed tests are measured in Kilobits /second.

Just to confuse matters, many internet programs measure speed in kB (kilobytes). A byte being 8bits.

Traditionally before windows came along most throughput was measured in (kilobits).
To confuse matters more, kilo can mean 1024 or 1000 depending on the scenario. In fact I'm almost certain I have mixed it up here explaining it to you.

As another poster said, just divide the score you get on the speedtest by 8 to get a rough estimate.

watty

x 10 as there is usually other overheads.

Memory chips have address lines. Each extra pin adds a power of 2 or 4. Hence a 10 address wired chip is 1024 locations. A 1M chip 1024 x 1204 locations.

But in everything else K = 1000, M = 1000 x 1000
Disks and data transfer tend to be measured "properly", i.e. multiples of 1000.

K and M on Memory should really be Kibi and Mibi etc..
http://en.wikipedia.org/wiki/Kibibit

watty

watty wrote: »

x 10 as there is usually other overheads.

Memory chips have address lines. Each extra pin adds a power of 2 or 4. Hence a 10 address wired chip is 1024 locations. A 1M chip 1024 x 1204 locations.

But in everything else K = 1000, M = 1000 x 1000
Disks and data transfer tend to be measured "properly", i.e. multiples of 1000.

K and M on Memory should really be Kibi and Mibi etc..
http://en.wikipedia.org/wiki/Kibibit

yes to summerise the people who thought of the two systems should be lined up and shot