Another Noob..kinda..

[krkaufman]

I found a thread on iPerf and it indicates "-P" (capital P) for parallel streams...

Test using more than one stream -- Use 5 parallel streams - I have found this to be a good number to use

iperf3 -c <server_ip> -P 5

I *believe* this is what provided the best test results for me.

 

[degrub]

Your bonded throughput rate on good RG6 should be at least 800 mbits/sec. physical rate should be up to 1200

start with post #46
https://www.snbforums.com/threads/moca-2-0-actiontech-ecb6000.26415/

 

[krkaufman]

And Actiontec shouldn't be abandoning you with the "your network" excuse if you can't get TURBO bonded 2.0 going between a pair of direct-connected ECB6200 adapters. Your network isn't a part of it if you're directly network two computers via the direct-connect setup.

 

[krkaufman]

Your bonded throughput rate on good RG6 should be at least 800 mbits/sec. physical rate should be up to 1200

Is there a post somewhere listing that as an observed PHY rate? I understand that bonded's PHY rate can range up to 1400 Mbps, but have only seen sub-700 Mbps numbers ... with the doubling coming from the bonded channel.

From the specs (well, a press release)...

MoCA, the standard for home entertainment networking, has ratified its next-generation specification, MoCA 2.0. MoCA 2.0 offers two performance modes, Basic and Enhanced, with 400 Mbps and 800 Mbps net throughputs (MAC), using 700 Mbps and 1.4 Gbps PHY rates, respectively. For point-to-point WAN applications, performance can be optimized for two node networks with Basic mode delivering 500 Mbps net throughput (MAC) and Enhanced mode providing 1 Gbps net throughput. (link)
​

p.s. Similar to what's reported in this post, and depicted in its attached bandwidth chart.

 

[cbelling]

I found a thread on iPerf and it indicates "-P" (capital P) for parallel streams...

​

I *believe* this is what provided the best test results for me.

Thanks. I ran that and the attached is what I got.

That's with PC Cat 6 -> Adapter->Coax->Adapter->Cat 6->Laptop.

 

[cbelling]

And Actiontec shouldn't be abandoning you with the "your network" excuse if you can't get TURBO bonded 2.0 going between a pair of direct-connected ECB6200 adapters. Your network isn't a part of it if you're directly network two computers via the direct-connect setup.

THANK YOU!!

 

[cbelling]

The laptop has a Gigabit Ethernet interface, apparently. And you've verified that the laptop is establishing a Gigabit Ethernet link to the MoCA adapter? And when doing the testing you're always disabling the laptop's wireless interface, just to make sure there's no confusion?


MoCA can work fine with most cable-rated splitters, but you're not just doing MoCA, now, you're pushing bonded MoCA 2.0 -- so where MoCA 1.1 just needed the cable splitter to not muck too much with signals in the 1125-1175 MHz range (the lowest 50 MHz-wide channel), bonded MoCA 2.0 requires 225 MHz of bandwidth. If having issues, it would be worthwhile to upgrade to known-good MoCA 2.0 splitters, either the Verizon MoCA 2.0 series you noted or Holland's GHS-PRO-M series (these).


That's pretty typical and shouldn't be a problem. If'/when your provider starts doing DOCSIS 3.1, you may find a need to run a dedicated coax line to the modem, but your current setup is pretty much what MoCA was designed to deal with.


You might try this test again, except you should be connecting the two MoCA adapters via each of their "COAX IN" ports. (I'm a little surprised the adapter was even able to make a connection, since the pass-through port severely attenuates the signal at MoCA frequencies.)

Oh, right, looks like you already did...

But what throughput were you seeing with the MoCA adapters direct-connected like this? (One connected via Ethernet directly to your router LAN, and the other to your laptop?)


Assuming you've found the magic iPerf command-line options, the test setup you just described, two ECB6200 MoCA adapters direct-connected, should be capable of TURBO mode bonded MoCA 2.0 ... up to 1000 Mbps (rather than the 800 Mbps top for non-TURBO bonded MoCA 2.0).

You're reporting PHY rates, so I'm guessing that you have access to the status page for the ECB6200 adapters. If so, can you post a screenshot of what you're seeing, including the frequency chart/diagram.

You are correct, Gigabit Ethernet link, and wireless turned off.

I've posted my results from the Actiontec recommended test.

 

[krkaufman]

Thanks. I ran that and the attached is what I got.

View attachment 14472

Isn't that iPerf test result showing a net throughput around 950 Mbps? (i.e. the "SUM" line, summarizing the bandwidth of each of the 5 parallel streams)

 

[degrub]

Is there a post somewhere listing that as an observed PHY rate? I understand that bonded's PHY rate can range up to 1400 Mbps, but have only seen sub-700 Mbps numbers ... with the doubling coming from the bonded channel.

From the specs (well, a press release)...

MoCA, the standard for home entertainment networking, has ratified its next-generation specification, MoCA 2.0. MoCA 2.0 offers two performance modes, Basic and Enhanced, with 400 Mbps and 800 Mbps net throughputs (MAC), using 700 Mbps and 1.4 Gbps PHY rates, respectively. For point-to-point WAN applications, performance can be optimized for two node networks with Basic mode delivering 500 Mbps net throughput (MAC) and Enhanced mode providing 1 Gbps net throughput. (link)
​

p.s. Similar to what's reported in this post, and depicted in its attached bandwidth chart.

Lab conditions, of course.

 

[cbelling]

Isn't that iPerf test result showing a net throughput around 950 Mbps? (i.e. the "SUM" line, summarizing the bandwidth of each of the 5 parallel streams)

I just discovered iPerf yesterday so if that's what you say it says then that is what it says . I guess then I'm getting the appropriate speeds now in closed loop conditions so I now need to actually bring the modem/router back into the mix?

 

[krkaufman]

I've posted my results from the Actiontec recommended test.

View attachment 14474

That looks really good, right... 670 Mbps PHY rate, and the bandwidth graph shows 2 bonded channels (so an effective PHY rate of 1340 Mbps) ... with very little variance.

 

[krkaufman]

Lab conditions, of course.

Right, totally agree.

edit: ... though I'd suggest "theoretical," since even a lab would likely have difficulty hitting the peak spec rates.

 

[krkaufman]

I just discovered iPerf yesterday so if that's what you say it says then that is what it says . I guess then I'm getting the appropriate speeds now in closed loop conditions so I now need to actually bring the modem/router back into the mix?

I like that the "SUM" results map to what we are expecting, so, yeah, I'm good with them, as well.

And, yes, go back to your regular setup and try iPerf again, using the "-P 5" option ... and look for the "SUM" results.

 

[cbelling]

I like that the "SUM" results map to what we are expecting, so, yeah, I'm good with them, as well.

And, yes, go back to your regular setup and try iPerf again, using the "-P 5" option ... and look for the "SUM" results.

Thanks. Gotta say, this is a fantastic forum. The help here is amazing.

 

[krkaufman]

I guess then I'm getting the appropriate speeds now in closed loop conditions ...

And those were TURBO/Enhanced bonded MoCA 2.0 speeds, so, yeah, thumbs up.

 

[cbelling]

After hooking up a dedicated cable, with regulation splitters I am getting a pretty damn good throughput of 1.09 gb.

The odd thing is my download speeds on my laptop are still a lot slower than my PC, but I can only attribute that to something on the laptop because in my little mini setup I plugged the PC into the router and got nearly 400mbps second, then I plugged it directly into the MoCA and got nearly the same result. (Did the same with the laptop and it was still a lot slower-100mbps).

 

[krkaufman]

After hooking up a dedicated cable, with regulation splitters I am getting a pretty damn good throughput of 1.09 gb.

That's what you're seeing, 1009 Mbps, as an iPerf test result (from the "SUM" line)?

I plugged the PC into the router and got nearly 400mbps

400 Mbps, but between what two points? Was this an Internet download test using something like speedtest.net? What download rate are you paying for?

The odd thing is my download speeds on my laptop are still a lot slower than my PC, but I can only attribute that to something on the laptop because in my little mini setup I plugged the PC into the router and got nearly 400mbps second, then I plugged it directly into the MoCA and got nearly the same result. (Did the same with the laptop and it was still a lot slower-100mbps).

In my searching, last night, I recall seeing someone reporting different iPerf results owing to one device being older/slower.

Oh, also, I just came across another post recommending some iPerf command-line options...

MoCA network has greater latency (~3.5 ms) than direct ethernet connection.

You should try to use iperf3 with options -w 1M -P 5.

I was using the default options for iperf3. Using iperf3 with the -w 1M option fixed it, and I am measuring >900 Mbps.

... though I don't understand the benefit of the "-w" option for MoCA:

-w, --window n[KM]
             window  size  / socket buffer size (this gets sent to the server
             and used on that side too)

Some learnin' to do.

-27 for the TX levels. I believe that's pretty good?

-30 dBm is the max theoretical power reduction for MoCA gear (+3 dBm being the max TX power amplification, where you'd begin seeing a drop-off in PHY rate), so, yeah, -27 is great ... and it's all I've even seen in any testing (i.e. I've never seen -30 dBm TX power). The lower power draw would indicate that the MoCA gear is not having to work that hard to overcome losses between nodes, so you shouldn't have to worry about upgrading splitters to MoCA 2.0, etc.

If the -27 dBm figure was for the direct-connect setup, you may want to take note of the PHY rate and power levels once the MoCA adapters are back in their permanent locations.

 

[cbelling]

That's what you're seeing, 1009 Mbps, as an iPerf test result (from the "SUM" line)?


400 Mbps, but between what two points? Was this an Internet download test using something like speedtest.net? What download rate are you paying for?

In my searching, last night, I recall seeing someone reporting different iPerf results owing to one device being older/slower.

Oh, also, I just came across another post recommending some iPerf command-line options...

​

... though I don't understand the benefit of the "-w" option for MoCA:

-w, --window n[KM]
             window  size  / socket buffer size (this gets sent to the server
             and used on that side too)

Some learnin' to do.


-30 dBm is the max theoretical power reduction for MoCA gear (+3 dBm being the max TX power amplification, where you'd begin seeing a drop-off in PHY rate), so, yeah, -27 is great ... and it's all I've even seen in any testing (i.e. I've never seen -30 dBm TX power). The lower power draw would indicate that the MoCA gear is not having to work that hard to overcome losses between nodes, so you shouldn't have to worry about upgrading splitters to MoCA 2.0, etc.

If the -27 dBm figure was for the direct-connect setup, you may want to take note of the PHY rate and power levels once the MoCA adapters are back in their permanent locations.

Yes, the 1.09 was my results. The 400mbps was done using Speedtest. I am paying for 600 mbps from my ISP so although it'd be nice to get 600, I'm happy with 400.