How to test speed of ethernet cable?

[DeadlyPilot]

So i found an ethernet cable and have no idea what category it is, is there a way i can test what speed its being run at? Linus said in one of his videos that your NIC will only go in full steps, from 1gbit down to 10mbit and never in between. So lets say i wanted to test if a cat5e cable was truly capable of 1gbps, how would i test that?

 

Hope this made sense. Thanks

 

Edit: **I meant 100mbps not 10

[Godlygamer23]

As this deals with networking, I have moved to the thread to Networking.

 

Please post in the proper sub-forum when forming topics.

[GDRRiley]

it should say on it but if it does not connect 2 PC together and see how fast it transfers. 10 100 1gb

[DeadlyPilot]

11 minutes ago, DeadlyPilot said:

So i found an ethernet cable and have no idea what category it is, is there a way i can test what speed its being run at? Linus said in one of his videos that your NIC will only go in full steps, from 1gbit down to 10mbit and never in between. So lets say i wanted to test if a cat5e cable was truly capable of 1gbps, how would i test that?

 

Hope this made sense. Thanks

so basically as long as it can transfer above 100mbps the cable should be capable of 1gbps?

[mariushm]

If you have no idea what category it is, chances are it's Cat5e which means it can do up to 1 gbps.

Cat5 or older cables would be maybe more than 10 years old nowadays, so I doubt you'd use a cable with the exterior insulation all messed up (which would happen in 10 years)

 

Look at the wires going into the connector. If the color order is the same in both connectors, you have a straight (normal) cable, if the color order is different in one jack you have a crossover cable (a cable designed specially to be used between two computers / two network cards)

Network cards capable of 1 gbps speeds are smart enough (and it's actually required by the 1gbps standard) to determine if the cable is crossover or normal and automatically reconfigure themselves to work right in any situation. So, if you use a crossover or normal cable, it doesn't matter, you'd still get up to 1 gbps speeds if both network cards (or your network card and your switch port) are 1gbps capable.

But the point is, few people buy crossover cables anymore because of this autodetection mechanism, so if your cable is crossover there's a high chance it's an old one, so it's probably only Cat5e or older.

 

For 1gbps speeds all four pairs of wires must be properly installed. For 100 mbps, just two pairs of wires are used and the other four wires  (2 pairs) are not used. The network cards would auto configure themselves to 1gbps and if there's some issues like for example one loose contact in the jack (and if you're lucky it's one of the four wires that are not in use when running at 100 mbps), then the network card would detect errors transmitting at 1 gbps and will fall back to running at 100 mbps.

 

Otherwise, it's really very hard to know the quality of a cable the way you think. There are benchmarks that can tell you the speed ... i haven't used any of these below... but LAN Speed Test and Netstress sound good (though they're not very scientific/ not very repeatable results) : http://www.smallnetbuilder.com/basics/lanwan-basics/31220-how-fast-is-your-network-five-ways-to-measure-network-speed

 

NTttcp seems more professional (but it's harder to use since it's command line only):    https://gallery.technet.microsoft.com/NTttcp-Version-528-Now-f8b12769

Techreport.com used this to test the Killer E2400 network card : http://techreport.com/review/29144/revisiting-the-killer-nic-eight-years-on/2

TomsHardware used Passmark PerformanceTest 8 to test powerline adapters : http://www.tomshardware.com/reviews/how-we-test-powerline-networking-adapters,4217.html#p5

 

There are more complex (and ridiculously expensive) hardware testing and benchmarking tools which can even simulate lost or corrupted packets of data just to see how well a network card would auto correct the data or how much time it takes to have data retransmitted and so on ... but i can't remember them off the top of my head.

 

[DeadlyPilot]

8 minutes ago, mariushm said:

If you have no idea what category it is, chances are it's Cat5e which means it can do up to 1 gbps.

Cat5 or older cables would be maybe more than 10 years old nowadays, so I doubt you'd use a cable with the exterior insulation all messed up (which would happen in 10 years)

 

Look at the wires going into the connector. If the color order is the same in both connectors, you have a straight (normal) cable, if the color order is different in one jack you have a crossover cable (a cable designed specially to be used between two computers / two network cards)

Network cards capable of 1 gbps speeds are smart enough (and it's actually required by the 1gbps standard) to determine if the cable is crossover or normal and automatically reconfigure themselves to work right in any situation. So, if you use a crossover or normal cable, it doesn't matter, you'd still get up to 1 gbps speeds if both network cards (or your network card and your switch port) are 1gbps capable.

But the point is, few people buy crossover cables anymore because of this autodetection mechanism, so if your cable is crossover there's a high chance it's an old one, so it's probably only Cat5e or older.

 

For 1gbps speeds all four pairs of wires must be properly installed. For 100 mbps, just two pairs of wires are used and the other four wires  (2 pairs) are not used. The network cards would auto configure themselves to 1gbps and if there's some issues like for example one loose contact in the jack (and if you're lucky it's one of the four wires that are not in use when running at 100 mbps), then the network card would detect errors transmitting at 1 gbps and will fall back to running at 100 mbps.

 

Otherwise, it's really very hard to know the quality of a cable the way you think. There are benchmarks that can tell you the speed ... i haven't used any of these below... but LAN Speed Test and Netstress sound good (though they're not very scientific/ not very repeatable results) : http://www.smallnetbuilder.com/basics/lanwan-basics/31220-how-fast-is-your-network-five-ways-to-measure-network-speed

 

NTttcp seems more professional (but it's harder to use since it's command line only):    https://gallery.technet.microsoft.com/NTttcp-Version-528-Now-f8b12769

Techreport.com used this to test the Killer E2400 network card : http://techreport.com/review/29144/revisiting-the-killer-nic-eight-years-on/2

TomsHardware used Passmark PerformanceTest 8 to test powerline adapters : http://www.tomshardware.com/reviews/how-we-test-powerline-networking-adapters,4217.html#p5

 

There are more complex (and ridiculously expensive) hardware testing and benchmarking tools which can even simulate lost or corrupted packets of data just to see how well a network card would auto correct the data or how much time it takes to have data retransmitted and so on ... but i can't remember them off the top of my head.

 

So a cable will ONLY run above 100mbps if its capable of running at a full 1gbps correct? 

 

also, 

It says its just a cat5 patch cable. Would it be possible to run into a cable that physically appeared to be a cat5e cable but for some reason wouldnt be able to reach 1gbps (assuming your within the 100m limit)? For example a cable that is just cheaply made or is a smaller gauge than needed, but claimed to be cat5e?

[mariushm]

It's possible, but unlikely.

 

Here's an example.

Some network cards and switches have "Green mode" or "eco power" or "power saving" or some other kind of power saving modes (usually under some kind of name, and which can be disabled from network card's driver settings in Device Manager).  My realtek network card has two such settings "Energy Efficient Ethernet" and "Green Ethernet" - i leave both enabled because the longest distance is a 3 meter patch cord between the network card and the router from my ISP so the settings make no difference in my case.

 

By default, these modes are often enabled and in the case of the smartest cards, this means that initially the network cards send data at full power and if they see that everything's fine, they gradually lower the power of their transmitters to save a few milliwatts of power. If they start to detect errors in transmission or reception, they go back a notch and raise the power.  Less smart network cards just use a lower power transmitter without so much analysis of the signal and just assume the cables are good enough.

 

There are network cables out there which are meant to be used as small patch cables, just a few meters long. These cheap network cables are made with CCA (copper clad aluminum), basically each wire in the network cable is made from several strands of aluminum, each covered in a very thin layer of copper. Unlike pure copper cables, these cables are cheaper to produce but the resistance of the wire is often higher.

 

For short distances the cable is advertised for (a few meters) this higher resistance of the wires is irrelevant, but if you were to buy and create a 40+ meters cable out of such type of cable, it could happen that in combination with the lower transmitter power of network cards with power savings mode enabled, there would be so many uncorrectable errors that the cards would think the cable is bad and would fall back to 100 mbps.

This actually happened, and it happens more often with even shorter cables if on the other end are cheap 5 or 8 port switches that have this "green" or power saving feature always on.

 

[gamerking]

1 hour ago, DeadlyPilot said:

So i found an ethernet cable and have no idea what category it is, is there a way i can test what speed its being run at? Linus said in one of his videos that your NIC will only go in full steps, from 1gbit down to 10mbit and never in between. So lets say i wanted to test if a cat5e cable was truly capable of 1gbps, how would i test that?

 

Hope this made sense. Thanks

 

Edit: **I meant 100mbps not 10

if you have a linux server or can boot one from a flash drive run iperf on the server side then run the client on another machine with the cable in the middle in question and ta da you got your speed