Speed Testing
 

At this moment I am experiencing a slowdown of the Internet and investigated the speed using Numion My test of choice but I also tested using Speednet VMs first choice. Numion gave me 3Gb or less and Speednet gave me a 10GB followed by a 7GB. My question is how does VM get Speednet to do it?

Re: Speed Testing
 

Better servers.

Re: Speed Testing
 

Try this site http://www.measurementlab.net/run-ndt

Re: Speed Testing
 

I wasn't impressed with that site.

Apart from not being able to copy & paste the result (makes it a load of rubbish actually), it gave me 3.49 meg up and 9.37 down.

Speedtest.net gave me 4.7 up and 50 down

Re: Speed Testing
 

Gives me the same result as speedtest.net every time.

Re: Speed Testing
 

Gave me a 91 down and 9.3 up which is the best speed test I've ever seen ..

Re: Speed Testing
 

Pretty much useless for me as well. Speedtest reports 33.8mbps down, that site reports 6.0mbps down. And on my 100/100 it reports 5.4mbps down and 78mbps up (while actually uploading at 93mbps)

Interestingly on my "other" line NDT reports "3.8E+2" mbps upload and "5.4E+2" download. Or in other words it's so useless it can't even display numbers correctly. That said the result itself turns out more accurate than speedtest.net - which reports 245mbps down and 57mbps up. So in fact the upload test seems better than Speedtest.net's, though speedtest.net's has always been utterly useless on fast lines.

Re: Speed Testing
 

Is this test suppose to stall at the preparing test... screen? Its been on it for ages

Re: Speed Testing
 

OK - I've been into this a bity more and have, of course, found the reporting buttons that allow me to copy & paste. Bearing in mind that Speedtest.net reports 50/4.7, this NDT test now reports 22.5/3.5 after tweaking up my RWIN buffer. This compares with my previous NDT result of 9.2/3.5.

The reports produced by NDT are quite deep (that's good), but leaves me with unanswerable questions. Here are some salient points.

Upload speed 3.5 mb/s
Download speed 22 mb/s
Network latency: 1.9e+2 msec round trip time

Jitter: 1.8e+2 msec
Your system: Windows Vista version 6.0
Java version: 1.6.0_26 (x86)

TCP receive window: 4156416 current, 4597504 maximum
9.9522E-5 packets lost during test
Round trip time: 173 msec (minimum), 350 msec (maximum), 188.55 msec (average)
Jitter: 177 msec
0 seconds spend waiting following a timeout
TCP time-out counter: 394
1685 selective acknowledgement packets received

No duplex mismatch condition was detected.
The test did not detect a cable fault.
Network congestion may be limiting the connection.
No network addess translation appliance was detected.

0.9382% of the time was not spent in a receiver limited or sender limited state.
3.74% of the time the connection is limited by the client machine's receive buffer.
Optimal receive buffer: 412876800 bytes
Bottleneck link: 10 Mbps Ethernet
0 duplicate ACKs set

What 10 meg bottleneck is that? My topology, punting the 50 meg service, is:

Modem 1Gig --> Router 200 meg (through 1 Gig port) --> Homeplug 200 meg (161 meg effective) --> Netgear GS108 Switch 1 Gig port to --> 1 Gig ethernet port on PC.

So there's something about this test and how it probes what that needs explaining to me at least.

The "Advanced" tab reports (amongst other things):

The theoretical network limit is 5.92 Mbps
The NDT server has a 1833.0 KByte buffer which limits the throughput to 151.97 Mbps
Your PC/Workstation has a 4489.0 KByte buffer which limits the throughput to 186.03 Mbps
The network based flow control limits the throughput to 83.00 Mbps
Client Data reports link is 'Ethernet',
Client Acks report link is 'Ethernet'
Server Data reports link is 'OC-48',
Server Acks report link is 'null'

So, at this stage, it merits further investigation and perhaps a bit of playing around with TCP Optimizer.

Re: Speed Testing
 

Ditto :confused:

Gave up with it in the end. :shrug:

Re: Speed Testing
 

It's displaying them fine, just in scientific notation :)

1e2 rep points deducted for lacking geekiness.

Re: Speed Testing
 

Lol, I know the 'E' notation, it's just pointless it takes more characters than just displaying the full number.

Scientific notation is for numbers that are too long to display comfortably otherwise, except in this case the SN is twice as long as the number itself. Why jump to scientific notation at 3 significant figures when it takes 6 characters to display it?

Re: Speed Testing
 

https://www.cableforum.co.uk/images/...2011/09/37.png

seems ok to me

---------- Post added at 12:16 ---------- Previous post was at 12:15 ----------

Your system: Mac OS X version 10.7.1
Java version: 1.6.0_26 (x86_64)

TCP receive window: 524280 current, 524280 maximum
5.1065E-5 packets lost during test
Round trip time: 19 msec (minimum), 103 msec (maximum), 40.16 msec (average)
Jitter: 84 msec
0 seconds spend waiting following a timeout
TCP time-out counter: 240
380 selective acknowledgement packets received

No duplex mismatch condition was detected.
The test did not detect a cable fault.
No network congestion was detected.
No network addess translation appliance was detected.

0.9876% of the time was not spent in a receiver limited or sender limited state.
0.0% of the time the connection is limited by the client machine's receive buffer.
Optimal receive buffer: 536862720 bytes
Bottleneck link: 45 Mbps T3/DS3 subnet
0 duplicate ACKs set

Re: Speed Testing
 

Hi

I am struggling to find out how to improve my connection speeds etc as it sometimes takes over 3 min to change a page on ebay.

As a novice I need to find a test program that informs whats what. Then I can post the results on here for some help.

Question: What program or test do I use as a starting point ?

Regards.

Re: Speed Testing
 

Masque suggested to try this site http://www.measurementlab.net/run-ndt