I have an MT6000 which was running 4.6.4 (non op24). I upgraded one of my LAN devices with 2.5G NIC and I noticed that I was not getting 2.5Gbps bidirectional speeds with iPerf
But the issue is that now 5Ghz wifi speeds are terrible with op24. (I was getting around 1400mbps on 4.6.4 non op24 and now less than 900mbps on 4.6.4 op24.
Question 1: Why are op24 wifi speeds so bad?
Question 2: Is it a known issue that the 4.6.4 non op24 firmware does not support fully bidirectional 2.5Gbps speeds?
#1, Some users (possibly in different countries) have reported that the WiFi speed of firmware op24 will be better than the stable. Please confirm if the WiFi configuration of the 2 different firmware, like same country code, same channel, same bandwidth, etc.
#2, Please try the Openspeedtest (or other speedtest services, Librespeedtest) to test again to double confirm, recommand to run the speedtest server in another local machine client device not the router)
#1, Some users (possibly in different countries) have reported that the WiFi speed of firmware op24 will be better than the stable. Please confirm if the WiFi configuration of the 2 different firmware, like same country code, same channel, same bandwidth, etc.
Wifi configuration is the same on both firmwares:
Same country
Channel Auto
80 Mhz
#2, Please try the Openspeedtest (or other speedtest services, Librespeedtest) to test again to double confirm, recommand to run the speedtest server in another local machine client device not the router)
For the wired speed tests, I do not have another 2.5Gbe device currently. Only the router and one PC.
For the Wireless speed tests I have also tried iperf3 between PC (2.5Gbe to router) <-> iPhone which used to top at about 1400/1500mbps on 4.6.4 non op24 but now tops to less than 900mbps too.
@WizMeister heh, I think we talked on Reddit. I see similar WiFi speeds on op24.
@MilesTEG I can't test this currently, because I am on op24 and my family wouldn't be happy if I yank the internet connection now, but for the NAS is sending case, did you have the chance to monitor CPU use? I wonder if it triggers some path that is not hardware-accelerated? (Just guessing here.)
You could see that the NAS is capable of sending near at 2,5Gbits/s as receiveing at the same speed.
For me, it's on the router there is a little problem.
For my use, it's not important because my internet provider gave me:
Download : 2,5Gbps on one ethernet port (really 5Gbps for all ethernet port and wifi)
Upload : 700 Mbps
So, this is my provider's connection that limits my ability to send on the internet.
The router may limits when I want to send through wifi to a NAS or NUC. But for me, that's more my macbook air that limits, I think.
The iperf3 results for my MBA on wifi, near the router (60 cm):
The thing is, it’s only iperf on the router. I get near 2500MBit up and down to the internet through the router (I’m on 4Gbit symmetric), also on the stable firmware (non-OpenWrt 24).
This is why I was wondering whether running iperf on the router (I didn’t want to imply your NAS couldn’t receive or send 2.5Gbit), is bypassing the hardware-acceleration somehow and why I asked what the CPU load is on the router in the lower-bandwidth scenario.
Flint 2
└ --- 2.5G WAN port --> Cellular router in ip passthrough mode (It has 1Gbe LAN port)
└ --- 2.5G LAN port --> Dell machine with 2.5G NIC.
I installed 4.7 beta to test the LAN speeds and it has the same problem as 4.6.4 op21. No bidirectional 2.5Gbe speeds. I then reinstalled 4.6.4-op24 and verified that I do have 2.5Gbe bidirectional speeds. So it has to do with 4.6.4-op21 and 4.7-op21 beta.
(For WiFi I believe it was my bad. I inspected the backup I had on my 4.6.4 op21 setup and observed that indeed I had 160Mhz DFS enabled.)
I am not reporting about speed tests over the internet so it shouldn't matter what my WAN connection is, I just mentioned it so you can have the full view of the topology. I am testing the speed between the router and the 2.5G LAN device.
Can the op version have 2Gb?
I do not understand your question. Which "op"? With 4.6.4-op24 I have bidirectional 2.5Gbe speeds as you see on the screenshot of my first post. With 4.6.4 non op-24 I do not get bidirectional 2.5Gbe. It is clearly shown on the screenshots I posted and @MilesTEG posted.
Doesn't matter who is the server, it matters the direction of the data. If the router is the receiver, then the speed is slow.
Example A:
Iperf server: Flint 2 (192.168.1.1)
Iperf client: Dell with 2.5Gbe NIC
If I run iperf3 -c 192.168.1.1 from the Dell machine (sender is Dell and receiver is Flint 2) then I have slow speed (about 1.13Gbps)
If I run iperf3 -c 192.168.1.1 -R from the Dell machine (reverse; sender is Flint 2 and receiver is Dell) then I have expected speed (about 2.3Gbps)
Example B:
Iperf server: Dell with 2.5Gbe NIC (192.168.1.69)
Iperf client: Flint 2
If I run iperf3 -c 192.168.1.69 from the Flint 2 (sender is Flint 2 and receiver is Dell) then I have expected speed (about 2.3Gbps)
If I run iperf3 -c 192.168.1.69 -R from the Flint 2 (reverse; sender is Dell and receiver is Flint 2) then I have slow speed (about 1.13Gbps)
sorry, by op I mean 4.6.4-op24
With 4.6.4-op24 I have consistent 2.5Gbps speeds on both examples.
I do not own another 2.5Gbe client device but I am expecting a USBC 2.5Gbe NIC the following days to use with my laptop. I will test and report within the week.
Probably MBA (Which generation?) wireless performance limits. Can press the keyboard 'option' and click the WIFI logo, to see the detail wireless connectivity info below the SSID.
