Buzz is spot on. The "speed" rate you are seeing in the adapter setting is only a PHY link rate and speaks to the connection rate and little to do with actual data throughput, wired or wireless. Jim's ipef/jperf recommendation is one that works well, two computers is a PITA but gives you better information and is free software. An old laptop is great to have for situations like these.

The tester Buzz links to is a quick way to get good data, albeit relatively expensive. There are a couple of classes of testers out there made by Fluke, Ideal, etc... and may be a worthy investment for some.

My2.

I can't figure out what to do here.

Reading the user docs, I can't figure out how to "basically set up a computer;" I guess you mean set up this computer as a server. I mean, I see command line options, but if you could give me some hints as to how to use these, I'd appreciate it.


Information says

So... these are commands I add to "iperf" on the command line, right? What does bold print mean? What does italics mean? How do I decide between the first example and the second example?


Since both computers should use the same server port, why is there an option to specify the client-side port? Seems like that would break the operation.

And so forth.

Nobody has a clue how to use this program?

Ernie, I just downloaded and had it running in about 3 minutes.

You need:
- 2 computers - Windows of any stripe or Mac - with their Ethernet ports connected at to either end of the cable to test (you'll o
- Both will need to have static IP addresses set on the same subnet (if you don't know how to do this, stop here and go buy or rent a certifier)
- Download the appropriate binaries for your operating system from iperf.fr, unzip or decompress into a temporary folder
- On each computer, open a command prompt (Windows) or terminal window (Mac), navigate to that temporary folder
- On one computer, type "iperf3 -s", that tells it to run in server mode. You'll see it respond "Server listening on port 5201".
- On the other computer, run "iperf3 -c XXX", where XXX is the IP address you set on the OTHER computer - that tells it to act as client, connect to the server, and run the test.

About 1 minute later the results will be displayed on the both the client and serve. For reference, my client is on a Gigabit wired link, my server is connected to that LAN over a 5 Ghz 802.11ac wifi connection showing a PHY rate of 867 Mbps (megabits/sec). Iperf is showing throughout of about 90 Mbps - a huge decrease because wifi connections have much higher overhead and never approach the physical link speed. If your cable allows a 100 Mbit connection speed I'd expect to see throughput of I'd guess 60-70 Mbps.

Well, thanks, but sure enough, this is still beyond me.
I don't know how to do step 2.
I don't know how to tell which are the appropriate binaries. This sentence even sounds like it's in code.

So how about a dumber method: would pinging another computer on the network be helpful?

Ernie,

PING can indicate the turnaround time for a few small packets that might be able to sneak through the clutter, but it is a very crude measure of throughput.

I doubt that you will run into this issue on your local network, but try PING cnn.com. Now, PING msn.com. !! Is msn.com down or unreachable from your location? Probably not, and you can verify this by using a browser to reach msn.com. My point is that PING is basically a "are you there, can you hear me?" probe, not a "can you understand me?" or "can we have a productive conversation?" measure (BTW: sites can specifically block PING)

If there is interference on the wired or wireless LAN, a fraction of the packets may be garbled and will need to be resent. It is not productive when you have a basic 100Mb or 1Gb raw connection if a significant percentage of packets need to be resent multiple times.

Sure, I use PING quite often. Since it is so easy to use, it is often my first test if I suspect trouble, but it is basically a check to see if things are plugged in. Note that one might have very poor PING times due to a packet "flood", but this is a configuration or basic compatibility issue, not a wire type problem. (Sonos users might bump into this after adding a mis-configured managed switch)

Going back to my earlier amplifier analogy, you can connect a pair of speakers and listen for noise, but if i cleverly pick the music, you will not notice gross frequency response problems or even clipping.  On my test bench these and other issues will be obvious after a minute or two of testing. True, if I want to go through the full FTC sequence, this will require at least an hour, because this is the the FTC test methodology. (pre-condition for an hour)

Bottom line: you may seem get away with the rude wiring that you are stuck with (or not), however, getting away with this today is a poor guarantee about tomorrow. Without proper testing, you are guessing. If you and the customer are comfortable with the risk -- go for it.

---

I have come across accounts of people using a couple pairs of zip-cord for their network making fun of the rest of us worrying over wire and wire routing. But, just as I'm sure that somewhere, someone is using Walmart's finest audio extension wire for their 50ft run of 1080p, I don't think that many of us would attempt that. And I have come across accounts of splicing HDMI and even extending HDMI using a CAT-5 splice. (I have not yet heard any boasts about a successful 18.5Gb HDMI CAT-5 splice) I am not about to waste my time fussing with such schemes.

You don't seem to be commenting on the scenario I described. These cables are not being used for anything AT ALL, so as I see it, there's no clutter.


...and/but does it give response times? Pinging a computer on a network from a computer on a network, with no other devices connected, could give worthwhile information.


As I said, I'd be wiring up whatever someone recommends in order to make this test. That is, there'd be nothing else on the network. Interference from where?


Packet "flood" from what? The only other device on the network?


Are simply out to be negative here? Who is going to cleverly choose situations with my two computers and nothing else on the network, so that I'd have similar errors?

I did not mention the Wifi situation because I would be turning it off, but what the heck, you might come up with a reason for THAT to cause a problem: the home is on the side of a hill and I get no other wifi signals with my Android wifi snooper.


You are exactly right. So far, the client has been an adult about the fact that he fell in love with the look of a flip. Here's hoping that there's nothing going on under the marble floor of the theater -- to which there is no access -- in the future!


And these guys are totally not recognizing that they're using the wrong approach. It's not necessary to rip apart twenty feet of wall to install speakers wires into a bathroom if, for some reason, there's a couple of CAT5s going there. These guys seem to think such wire should be specified as the goal, rather than using it despite its limitations.


Your attitude is trumping your sense here. What Walmart audio cable has enough conductors to run 1080P? Isn't HDMI required for 1080P, so a PAIR of audio cables could not possibly do it?


You're not specifying here whether you mean actual HDMI with its 19 wires, or "HDMI over CAT," which is NOT HDMI, even though it might pass a signal equal in performance to HDMI.

You are correct not to waste your time on this. But the cases I've heard of have been installed wire with rather large replacement costs. Again, it's like using CAT wire for speaker wire: if you MUST do it, there will be limitations but it can do something. It should never be specced as something to install. Digital signals have more problems than analog in this realm of usage since analog can be lowered in quality but there's no "cliff" past which you'll get a signal, but an unusable signal, as with digital signals.

Ernie, the right way to do this is with a cable qualifier or certified. The former will tell you what data rate the cable SHOULD be able to support, the latter actively tests to validate the data rates (and in some cases network protocols) the cable DOES support.

You can do a reasonable approximation of these tests using PCs - but you have to be able to get them connected across the cable and set IP addresses and subnets manually on both so they can communicate with one another. As pointed out above, if you don't know how to do that basic setup, you're in over your head - either break out the books or hire a pro. And sorry if that's blunt, but if you happened on someone in the tool corral at Home Depot wondering out loud which way the blade goes into the Sawzall, you'd have the same response.

Assuming you get them up and communicating, the only positive test is a throughput test, which times how long it takes to push a bunch of data through. Errors on the link will show in the form of lower-than-expected throughput. iperf is a very good tool for doing that, and the instructions I provided above are right at the level for someone who knows how to get the Sawzall plugged in and the right blade selected and properly mounted.

Speedtest is another version of a throughput test, but because the test server is hosted on the internet, you'd need one end of your cable connected to a route and out to internet to use it. And, because their internet connection is probably the slowest link in the chain, and there are many, many variables from the router out to the internet, at best this will only show you that the cable works reliably at the speed of their internet connection, which may be a fraction of the data rate you'd like the cable to support.

Ping (which BTW won't work until you pass the connectivity prerequisites above!) is a latency, not throughput test. It tells you how long it takes very small packets to be sent back and forth. Most useful for identifying congestion or loss on a multi-hop network route. Even if ping is reliably sub-millisecond and lossless (indicating it's working OK at the physical link rate at which the devices are connected), it's no guarantee the link will continue to work well when trying to push lots of data through at that same rate. Of course, if it does show high round trip times, almost certainly caused by losses (since there is no other traffic on the link you're testing) it DOES mean there is a problem at that physical link rate... you could then set a slower rate on both PCs (assuming you know how to swap blades on the Sawzall) and try again, but even if/when it works at 10 or 100 Mbps you still won't have proved the cable can reliably sustain that data rate.

Ernie,

It would seem that PING and Speedtest are the only tools in your toolbox. For your purposes Speedtest is the best of the lot.

Guys,
thanks for helping with my education. I'm late to this party. And bluntness is not nearly as bad as pussyfooting around and not clearly stating the situation!
Thanks, again!