Actiontec ECB6200 Bonded Speed

[languagegame]

I have two Actiontec ECB6200 adapters, and I am testing performance using iperf3 between two Windows 10 PCs. The highest performance that I've been able to measure using the Moca adapters is around ~380 Mbps between the PCs, which is great but slower than I was expecting. I can get >850 Mbps using ethernet only, and I was expecting 800-900 Mbps with the Moca adapters based on reading reviews and other forum posts. Is this incorrect or should I be expecting higher than I am measuring?

Here is my setup:
- The two Moca adapters are connected directly using the included coax cable from the box, which is around 3 ft. No coax splitter is involved, and the PCs are connected to each adapter using the included ethernet cables.
- The adapters have SW firmware 2.11.1.50.6200.7, board ID 6802ECB, and they are configured using the default settings (Moca privacy = disabled, RF Channel = 1450, RF Band = Extended D and RF Switch = Hi)
- The node info page shows a RX Phy Rate of 670 Mbps, which seems consistent with the ~380 Mbps throughput measured using iperf3 assuming Phy overhead. The SNR db is 37, and RX and TX Power Db are -35 and -27 respectively.

I don't think the PCs are a bottleneck, as I can connect them directly using ethernet and measure >850Mbps (bypassing the Moca adapters entirely). The speed drops to ~380 Mbps when I add the moca adapters to the chain (PC -> ethernet -> moca -> moca -> ethernet -> PC).

Questions:
- Should I be expecting higher throughput using iperf3 than 380 Mbps with these Moca 2.0 bonded adapters?
- Do you think that the adapters might be using only a single channel vs. bonded? That may explain the slower speed, but I did not see a way to check or enable/disable channel bonding?
- Do you think the adapters might have fallen back to using Moca 1.1 vs. 2.0, which might explain the lower than expected speed? Is there a way to force Moca 2.0?

Some observations:
- I tried various RF channels in the Extended D band, and I get roughly the same performance around ~380 Mbps. I also tried the D Hi band, and got the same result.
- I tried the E and F bands, and while the settings would save to the adapters, the Coax link would not light up and no connection was possible. This seems consistent with the product documentation that says that the device supports Extended D only (which includes the D Hi band)

 

[degrub]

I have two Actiontec ECB6200 adapters, and I am testing performance using iperf3 between two Windows 10 PCs. The highest performance that I've been able to measure using the Moca adapters is around ~380 Mbps between the PCs, which is great but slower than I was expecting. I can get >850 Mbps using ethernet only, and I was expecting 800-900 Mbps with the Moca adapters based on reading reviews and other forum posts. Is this incorrect or should I be expecting higher than I am measuring?

Here is my setup:
- The two Moca adapters are connected directly using the included coax cable from the box, which is around 3 ft. No coax splitter is involved, and the PCs are connected to each adapter using the included ethernet cables.
- The adapters have SW firmware 2.11.1.50.6200.7, board ID 6802ECB, and they are configured using the default settings (Moca privacy = disabled, RF Channel = 1450, RF Band = Extended D and RF Switch = Hi)
- The node info page shows a RX Phy Rate of 670 Mbps, which seems consistent with the ~380 Mbps throughput measured using iperf3 assuming Phy overhead. The SNR db is 37, and RX and TX Power Db are -35 and -27 respectively.

I don't think the PCs are a bottleneck, as I can connect them directly using ethernet and measure >850Mbps (bypassing the Moca adapters entirely). The speed drops to ~380 Mbps when I add the moca adapters to the chain (PC -> ethernet -> moca -> moca -> ethernet -> PC).

Questions:
- Should I be expecting higher throughput using iperf3 than 380 Mbps with these Moca 2.0 bonded adapters?
- Do you think that the adapters might be using only a single channel vs. bonded? That may explain the slower speed, but I did not see a way to check or enable/disable channel bonding?
- Do you think the adapters might have fallen back to using Moca 1.1 vs. 2.0, which might explain the lower than expected speed? Is there a way to force Moca 2.0?

Some observations:
- I tried various RF channels in the Extended D band, and I get roughly the same performance around ~380 Mbps. I also tried the D Hi band, and got the same result.
- I tried the E and F bands, and while the settings would save to the adapters, the Coax link would not light up and no connection was possible. This seems consistent with the product documentation that says that the device supports Extended D only (which includes the D Hi band)

Weak RG 6 cable ? too short ?

 

[abailey]

It is not MOCA 1.1 as that cannot run at 380Mb/s. It sounds a lot like MOCA 2.0 non bonded which has a max (non-turbo) rate of 400Mb/s. Very close to your 380. As to why your getting the reduced speed, I am not sure. I would try a longer piece of COAX. I have heard of problems when using an extremely short piece of COAX, like 3 feet.

 

[uranus]

I don't think the PCs are a bottleneck, as I can connect them directly using ethernet and measure >850Mbps (bypassing the Moca adapters entirely). The speed drops to ~380 Mbps when I add the moca adapters to the chain (PC -> ethernet -> moca -> moca -> ethernet -> PC).

Questions:
- Should I be expecting higher throughput using iperf3 than 380 Mbps with these Moca 2.0 bonded adapters?

MoCA network has greater latency (~3.5 ms) than direct ethernet connection.
You should try to use iperf3 with options -w 1M -P 5.

 

[languagegame]

MoCA network has greater latency (~3.5 ms) than direct ethernet connection.
You should try to use iperf3 with options -w 1M -P 5.

Wow. Thanks everyone. I was using the default options for iperf3. Using iperf3 with the -w 1M option fixed it, and I am measuring >900 Mbps. This works both with the included 3 ft Coax cable or using my home's in-wall wiring (much longer than 3 ft).

 

[sfx2000]

Cool...

More iperf3 fun...

iperf3 -P 10 -w 1M -Z m -t 10 -i 1 -V -c iperf.he.net -R

This tests downlink from the source to the sink, to test the other way - remove the -R switch

 

[sfx2000]

And for even more fun with iperf3...

sfx@blaster:~$ iperf3 -P 10 -w 1M -Z m -t 10 -i 1 -u -b 10000M -c 192.168.1.20 | grep SUM
[SUM]   0.00-1.00   sec   478 MBytes  4.01 Gbits/sec  61180 
[SUM]   1.00-2.00   sec   484 MBytes  4.06 Gbits/sec  62000 
[SUM]   2.00-3.00   sec   489 MBytes  4.10 Gbits/sec  62550 
[SUM]   3.00-4.00   sec   486 MBytes  4.08 Gbits/sec  62270 
[SUM]   4.00-5.00   sec   488 MBytes  4.10 Gbits/sec  62490 
[SUM]   5.00-6.00   sec   488 MBytes  4.09 Gbits/sec  62410 
[SUM]   6.00-7.00   sec   489 MBytes  4.10 Gbits/sec  62550 
[SUM]   7.00-8.00   sec   488 MBytes  4.09 Gbits/sec  62480 
[SUM]   8.00-9.00   sec   488 MBytes  4.10 Gbits/sec  62510 
[SUM]   9.00-10.00  sec   487 MBytes  4.09 Gbits/sec  62360 
[SUM]   0.00-10.00  sec  4.75 GBytes  4.08 Gbits/sec  0.002 ms  0/622800 (0%)

 

[Samir]

Thank you for doing this! I was wondering how good these were out of the box. I've got one leg of our network served by powerline that I was thinking of converting to Moca--and these results definitely show that it would be a vast improvement for sure.

 

[Douchey]

Thanks for this information. This also solved my problem. I was testing with iperf and getting the same speed as the OP. I was banging my head against the wall, as I have all the right splitters, and f-connectors and cables. I couldn't figure out why I wasn't getting the full two channel speed. Running iperf with the "-w 1M -P 5" parameters allowed me to get the full speed.