For someone that might purchase a corporate MBA or MBP, is this issue only with USB-C adapters? Or is also with USB-C docking stations that contains ethernet? We use Dell d6000s docking stations that contains ethernet, and then connects to that docking station with USB-C to the laptop. Would this setup also have issues if I get a MBA or MBP?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
USB-C Ethernet unreliable
- Thread starter mikeboss
- Start date
- Sort by reaction score
The docking stations with Ethernet feature USB Ethernet adapters internally. So it’s the same issue. Gigabit ports will be unreliable. 2.5G ports will be much better.
As a follow up to my previous post, I have completed my comparison i) Belkin 1 Gbps adapter ii) D-Link DUB E-250 2.5 Gbps adapter iii) Plugable USBCE-25000 2.5 Gbps adapter and iv) UGreen 2502 2.5 Gbps adapter.
As far as the test environment and testing are concerned:
- I have a 1.5 Gbps fibre internet connection which when tested at the modem gave upload / download speeds of 940 Mbps / Mbps 940 when connected with an ethernet cable (and the D-Link adapter) to a 1Gbps port on the modem. The testing speeds are therefore relative to 940 Mbps / 940 Mbps as this was the best / max connected directly to the modem (and represents typical performance).
- I have modem connected to a Luxul Epic 5 1 Gbps router which in turn is connected to a number of Luxul AGS-1024 1 Gbps switches. The testing was limited to a maximum speed of 1Gbps because of the router's and swutches' limitations.
- Testing was conducted using the Okla Speedtest app running on a 2021 M1 Max MacBook Pro and an Apple Studio Display. The testing used the same endpoints, private relay off (i.e., the reason for using the app as private relay severely reduces download speeds), and wifi off. The testing was the best possible though not perfect (i.e., time of day, endpoint consistency, endpoint selection, local network traffic, etc.) though there were some clear trends.
- The testing compared the compared the download speeds and uploads with the ethernet connected i) directly to the MacBook Pro and ii) directly to the ASD and daisy chained to the MBP. The testing was performed for each of the four adapters.
- The testing measured the download / upload speed 15 different times for each adapter. The results were then used to calculate i) the average (faster is better) ii) the max - min (lower / smaller is better) and iii) the standard deviation (lower / smaller is better).
- The test results were the ranked and averaged to determine the best ethernet adapters.
As far as the speed test test results are concerned:
- The raw data (i.e., actual runs) and analysis are attached this posting.
As far as the analysis is concerned, some of the key observations are:
- Belkin Adapter: The reduction in download / upload speeds between connecting the ethernet cable directly to the MacBook versus connecting the ethernet cable to the ASD -> MacBook is real. The speed reduction averaged 295 Mbps / 190 Mbps. No other adapter exhibited a significant reduction in download / upload speeds.
The Belkin adapter was the only adapter that use the Realtek 8153 chip and the com.apple.DriverKit.AppleUserECm drive. The other adapters use the Realtek 8156 chip and the com.apple.driver.usb.cdc.ncm driver (with this also being the driver that the ASD uses!). Whether one or both of these differs explains the speed reduction is for others to answer (i.e., it is beyond my capability).
- D-Link Adapter: This adapter had the fastest download speed regardless of connection, the fastest upload speeds averaged between the connection, and a number of other good test results (i.e., minimal speed loss when concencting through the ASD, small standard deviation, etc.).
This adapter would have been the clear winner as it regularly produced trial run download speeds > 910 Mbps (relative to 940, good job Luxul) save for one very bad run in the 750 Mbps range.
This adapter also feel premium as D-Link added a number of "fine touches"; for example, the adapter is listed as DUB250E rather than teh generic 10/100/1000/2.5. This is a feeling of attention to detail.
This adapter was impossible to find new, no matter where I looked. I picked up a Certified Refurbed unit from D-Link and it looks / seems fine.
- Pluggable Adapter: This adapter performed very closely / similarly to the D-Link adapter especially when one compares the performance when teh ethernet cable is connected to the ASD -> MacBook with the two adapters testing very close to each other.
This adapter also comes with great technical support. This is second pluggable product I have owned and will give a shout out for their tech support; it is US based, knowledgeable and fast. While I did not think I would like the e-mail / forum only communication, it works very well.
-UGreen Adapter: This adapter worked (i.e., there was minimal download speed loss when connected through the ASD, there was almost no upload speed loss when connected through the ASD) , I want to stress this point. The results on average (when looking across all categories / test) were more variable (i.e., higher min / max difference and higher standard deviations) than the D-Link and the Plugable adapter.
Were this the only adapter that I ordered, I would have been thrilled because it solves the speed loss associated with the Belkin adapter.
- Caveat: The testing was limited to the four listed adapters (i.e., there are likely others that are as good and possibly better). The testing was not perfect and could be influenced by the time of day, the endpoint consistency, endpoint selection, local network traffic, etc.
Conclusions
- The D-Link and the Plugable adapter both solve the Belkin speed problem when connected to the ASD with aplomb (i.e., minimal variance when connected through the ASD (as compared to a direct MacBook connection).
- The D-Link and the Plugable when comparing and ranking only the D-Link, Plugable, and UGreen adapters (i.e., Belkin should not be a consideration because of the speed loss when connected through the ASD though it perform extremely well when connected directly to the MacBook) tied based on my ranking system.
- What adapter to keep? I am leaning towards i) keeping the D-Link (and return the other two) or ii) keeping the D-Link for home use, the Plugable for travel and returning the UGreen. What do you think I should do noting other options / suggestions are welcome?
I hope this was helpful and interesting.
I welcome comments from others.
Thank you.
As far as the test environment and testing are concerned:
- I have a 1.5 Gbps fibre internet connection which when tested at the modem gave upload / download speeds of 940 Mbps / Mbps 940 when connected with an ethernet cable (and the D-Link adapter) to a 1Gbps port on the modem. The testing speeds are therefore relative to 940 Mbps / 940 Mbps as this was the best / max connected directly to the modem (and represents typical performance).
- I have modem connected to a Luxul Epic 5 1 Gbps router which in turn is connected to a number of Luxul AGS-1024 1 Gbps switches. The testing was limited to a maximum speed of 1Gbps because of the router's and swutches' limitations.
- Testing was conducted using the Okla Speedtest app running on a 2021 M1 Max MacBook Pro and an Apple Studio Display. The testing used the same endpoints, private relay off (i.e., the reason for using the app as private relay severely reduces download speeds), and wifi off. The testing was the best possible though not perfect (i.e., time of day, endpoint consistency, endpoint selection, local network traffic, etc.) though there were some clear trends.
- The testing compared the compared the download speeds and uploads with the ethernet connected i) directly to the MacBook Pro and ii) directly to the ASD and daisy chained to the MBP. The testing was performed for each of the four adapters.
- The testing measured the download / upload speed 15 different times for each adapter. The results were then used to calculate i) the average (faster is better) ii) the max - min (lower / smaller is better) and iii) the standard deviation (lower / smaller is better).
- The test results were the ranked and averaged to determine the best ethernet adapters.
As far as the speed test test results are concerned:
- The raw data (i.e., actual runs) and analysis are attached this posting.
As far as the analysis is concerned, some of the key observations are:
- Belkin Adapter: The reduction in download / upload speeds between connecting the ethernet cable directly to the MacBook versus connecting the ethernet cable to the ASD -> MacBook is real. The speed reduction averaged 295 Mbps / 190 Mbps. No other adapter exhibited a significant reduction in download / upload speeds.
The Belkin adapter was the only adapter that use the Realtek 8153 chip and the com.apple.DriverKit.AppleUserECm drive. The other adapters use the Realtek 8156 chip and the com.apple.driver.usb.cdc.ncm driver (with this also being the driver that the ASD uses!). Whether one or both of these differs explains the speed reduction is for others to answer (i.e., it is beyond my capability).
- D-Link Adapter: This adapter had the fastest download speed regardless of connection, the fastest upload speeds averaged between the connection, and a number of other good test results (i.e., minimal speed loss when concencting through the ASD, small standard deviation, etc.).
This adapter would have been the clear winner as it regularly produced trial run download speeds > 910 Mbps (relative to 940, good job Luxul) save for one very bad run in the 750 Mbps range.
This adapter also feel premium as D-Link added a number of "fine touches"; for example, the adapter is listed as DUB250E rather than teh generic 10/100/1000/2.5. This is a feeling of attention to detail.
This adapter was impossible to find new, no matter where I looked. I picked up a Certified Refurbed unit from D-Link and it looks / seems fine.
- Pluggable Adapter: This adapter performed very closely / similarly to the D-Link adapter especially when one compares the performance when teh ethernet cable is connected to the ASD -> MacBook with the two adapters testing very close to each other.
This adapter also comes with great technical support. This is second pluggable product I have owned and will give a shout out for their tech support; it is US based, knowledgeable and fast. While I did not think I would like the e-mail / forum only communication, it works very well.
-UGreen Adapter: This adapter worked (i.e., there was minimal download speed loss when connected through the ASD, there was almost no upload speed loss when connected through the ASD) , I want to stress this point. The results on average (when looking across all categories / test) were more variable (i.e., higher min / max difference and higher standard deviations) than the D-Link and the Plugable adapter.
Were this the only adapter that I ordered, I would have been thrilled because it solves the speed loss associated with the Belkin adapter.
- Caveat: The testing was limited to the four listed adapters (i.e., there are likely others that are as good and possibly better). The testing was not perfect and could be influenced by the time of day, the endpoint consistency, endpoint selection, local network traffic, etc.
Conclusions
- The D-Link and the Plugable adapter both solve the Belkin speed problem when connected to the ASD with aplomb (i.e., minimal variance when connected through the ASD (as compared to a direct MacBook connection).
- The D-Link and the Plugable when comparing and ranking only the D-Link, Plugable, and UGreen adapters (i.e., Belkin should not be a consideration because of the speed loss when connected through the ASD though it perform extremely well when connected directly to the MacBook) tied based on my ranking system.
- What adapter to keep? I am leaning towards i) keeping the D-Link (and return the other two) or ii) keeping the D-Link for home use, the Plugable for travel and returning the UGreen. What do you think I should do noting other options / suggestions are welcome?
***
I hope this was helpful and interesting.
I welcome comments from others.
Thank you.
Last edited:
Thanks very much. Do you also see the numerous netstat output errors with the NCM driver, as reported in my last post, with your two 8156 based adapters? Do you notice any packet loss? I‘m torn between a rock and a hard place: The 8156 adapters are more efficient, but the 8153’s don‘t report any errors.
I am opting to stick with the 8156 based adapters as they provide the speed that I am looking forward and, while not measured, I have never / not yet noticed any missing or lost information.
In addition, I did not observe / test for that but would be please to do so should you wish.
Would you be so kind to test the terminal command "netstat -di -I enX" (use the numbered interface corresponding to your USB-C-Ethernet adapter) and report the number of "output errors" after some time of usage?
I would be happy to but want to make sure I do it correctly.
Is my understanding correct that based on the below image teh command should be netstat -di -I en4
Please correct / confirm my understanding at which point I will run the command and report back.
Thank you.
It reports, like on my M3 MBA, apparently lots of "Oerrs" (output errors) and some dropped frames. Using 8153-based adapters or the official Apple Thunderbolt Gigabit Ethernet adapter, there usually are zero "Oerrs".
@learjet I appreciate your explanation, it is helpful to me.
I will keep the Apple Thunderbolt Gigabit adapter just inc case but I am sticking with the D-Link / Plugable as i) I value teh speed and ii) I have noticed any data loss.
Thank you.
I just tested an Anker USB-A GB Ethernet adapter and it connected but I didn't do any speed testing. I bought this for my 2014 MacBook Pro 15 back around then and never had any problems with it with the old Intel MacBook Pros. I also have a few Thunderbolt to GB Ethernet adapters and I'd like to test that but I'd need a TB3 to TB2 adapter. I rarely use Ethernet with my MacBook Pro though I have it running with my desktops. I suppose that there is always the case where I might need to access it at a hotel or office.
Last edited:
I have terrible problems with the usb-c to ethernet Belkin F2CU040.
I will order another adaptor immediately for my m2 max macbook pro.
I will order another adaptor immediately for my m2 max macbook pro.
I had endless problems with my M1 Max and my NAS drives (a Qnap and Synology). They're frequently lose connection, especially when transferring files. This occurred with both a cheap 2.5g USB adapter and also the 10G one in OWC dock. Despite loving the machine, I was considering moving to another system due to the constant disconnects.
Anyway, upgrading to Sequoia I'm delighted to say completely fixed these problem and my NAS drives just work now as they should.
Anyway, upgrading to Sequoia I'm delighted to say completely fixed these problem and my NAS drives just work now as they should.
@JoelBC The problem with the Belkin F2CU040 model is not only about speed.
The adapter has very serious problems on a silicon macbook pro that are solved only me restarting (!) the laptop.
One easy way to understand that your adaptor has serious problems is that the network tab at the settings takes some time to appear!
The adapter has very serious problems on a silicon macbook pro that are solved only me restarting (!) the laptop.
One easy way to understand that your adaptor has serious problems is that the network tab at the settings takes some time to appear!
That Belkin is based on an 8153 chipset from Realtek. it has issues on Mac. See this and for more details see the link within that post: https://forums.macrumors.com/threads/usb-c-ethernet-unreliable.2287743/post-30886809
Yes I have seen that.
To be honest I never thought that the problematic part of the network line would be this specific part.
So I bought yesterday another switch and another ethernet cable to replace the old ones and then when the problem continued to exist, I tried to investigate the usb-c to ethernet adaptor case.
In fact thankfully I have an xbox also connected to the switch so when the macbook pro was unable to reach the router I immediately turned on the xbox and it had no problems at all connecting to the internet.
I bought D-Link DUB E-250 2.5 Gbps adapter.
No problems so far!
No problems so far!
Still zero problems with the D-Link adapter after 10 days.
I believe I can safely say now that it is rock solid.
I believe I can safely say now that it is rock solid.
Today I had some connection problems but after turning on the wifi the problems were still there, so it was a matter of the operating system or some 3rd party software probably, because a restart fixed the things.
Based on the advice in this thread, I picked up the D-Link DUB-E250 on sale here. I don't know what the usual pricing is but list price on the D-Link Canada website is CA$44.99. However, they had it on sale for CA$19.99.* I guess that moved a bunch of them quickly because now it's sold out.
I don't actually need it for my M4 Mac mini, since I got the 10 GbE upgrade on it. However, I did want to have 2.5 GbE support on a second machine for testing purposes, as I will be upgrading parts of my home network to 2.5 Gbps. The plan was to use my 2017 MacBook, but clearly it is not up to the task. The adapter shows up on the MacBook as connected at 2500 Mbps, but the MacBook just can't keep up.
So obviously this 12" MacBook will be useless for testing out my network. The USB-C port is only 5 Gbps but that should be fast enough for 2.5 GbE, but I also see that it's eating up a lot of CPU cycles, so perhaps that's where the bottleneck resides.
However, it turns out my M4 iPad Pro works great with it. I'm on 1.5 Gbps download with 50 Mbps upload service, and the iPad Pro is more than maxing that out.
*If it does get back in stock, here is a hint for a discount. At least in Canada, for purchases this small, they charge a shipping fee. However, D-Link Canada has a Subscribe And Save deal for new subscribers. Unfortunately, it's only 10% off, but more importantly, there is also a promo code floating around out there that provides for free shipping for subscribed users. You can't use the 10% discount and free shipping at the same time, but the free shipping is a much higher value. I subscribed to their news info, and did NOT get the discount code email as promised, but I found both the discount code and the free shipping promo code online anyway and was able to use the latter, so I just paid the CA$19.99 (which is a little over US$14) plus tax. (The codes did not work with my other email. They only worked with the email I had subscribed with.) At this price, it's a steal, significantly cheaper than the UGREEN for example.
I don't actually need it for my M4 Mac mini, since I got the 10 GbE upgrade on it. However, I did want to have 2.5 GbE support on a second machine for testing purposes, as I will be upgrading parts of my home network to 2.5 Gbps. The plan was to use my 2017 MacBook, but clearly it is not up to the task. The adapter shows up on the MacBook as connected at 2500 Mbps, but the MacBook just can't keep up.
So obviously this 12" MacBook will be useless for testing out my network. The USB-C port is only 5 Gbps but that should be fast enough for 2.5 GbE, but I also see that it's eating up a lot of CPU cycles, so perhaps that's where the bottleneck resides.
However, it turns out my M4 iPad Pro works great with it. I'm on 1.5 Gbps download with 50 Mbps upload service, and the iPad Pro is more than maxing that out.
*If it does get back in stock, here is a hint for a discount. At least in Canada, for purchases this small, they charge a shipping fee. However, D-Link Canada has a Subscribe And Save deal for new subscribers. Unfortunately, it's only 10% off, but more importantly, there is also a promo code floating around out there that provides for free shipping for subscribed users. You can't use the 10% discount and free shipping at the same time, but the free shipping is a much higher value. I subscribed to their news info, and did NOT get the discount code email as promised, but I found both the discount code and the free shipping promo code online anyway and was able to use the latter, so I just paid the CA$19.99 (which is a little over US$14) plus tax. (The codes did not work with my other email. They only worked with the email I had subscribed with.) At this price, it's a steal, significantly cheaper than the UGREEN for example.
This is actually a (relatively) known issue with Sequoia, see https://forums.macrumors.com/thread...on-a-2g-plan-since-upgrading-to-15-0.2437227/ or https://arstechnica.com/civis/threads/sequoia-2-5gbit-ethernet-slower-than-sonoma.1503073/.
My ISP is 2500/500 and this is what I get if I run the speedtest on my NAS, while my Mac Mini with a 2.5 Gbit adapter only gets this. They used to be basically the same before Sequoia came along, and it's still not fixed as of 15.2.
No, my MacBook is a 2017 model with slow CPU, and the max supported macOS version is Ventura. I can't even hit 550 Mbps with that D-Link 2.5 Gbps adapter, in Ventura.
However...
I guess that could explain why my 2.5 Gbps adapter maxes out at 1.95 Gbps in iPerf.
Server is M4 Mac mini with Sequoia 15.1.1, with 10 GbE port. Client is M4 iPad Pro with iPadOS 18.1.1, with D-Link 2.5 GbE adapter. They are connected through the same 2.5 GbE switch.
Testing with my internet access isn't helpful, because it is 1.5 Gbps service, which maxes out at 1.9 Gbps (which is apparently the speed to which my internet service is over-provisioned).
Last edited:
My bad, I had missed that part! An older machine may definitely be constrained by the transfer.
Regarding the newer ones (and the iPad, whose latest iPadOS also seems to be affected by the same issue as the latest macOS), you may find that upload speeds are not limited.
Between my Mac Mini and my NAS, this is what I get if the Mac is receiving:
Code:
octavarium:~ jollino$ iperf3 -c 192.168.0.50 -R
Connecting to host 192.168.0.50, port 5201
Reverse mode, remote host 192.168.0.50 is sending
[ 5] local 192.168.0.2 port 50238 connected to 192.168.0.50 port 5201
[ ID] Interval Transfer Bitrate
[ 5] 0.00-1.00 sec 232 MBytes 1.93 Gbits/sec
[ 5] 1.00-2.00 sec 232 MBytes 1.95 Gbits/sec
[ 5] 2.00-3.00 sec 231 MBytes 1.95 Gbits/sec
[ 5] 3.00-4.00 sec 233 MBytes 1.95 Gbits/sec
[ 5] 4.00-5.00 sec 232 MBytes 1.95 Gbits/sec
[ 5] 5.00-6.01 sec 232 MBytes 1.95 Gbits/sec
[ 5] 6.01-7.01 sec 232 MBytes 1.95 Gbits/sec
[ 5] 7.01-8.01 sec 232 MBytes 1.95 Gbits/sec
[ 5] 8.01-9.00 sec 232 MBytes 1.95 Gbits/sec
[ 5] 9.00-10.01 sec 232 MBytes 1.95 Gbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-10.01 sec 2.27 GBytes 1.95 Gbits/sec 203 sender
[ 5] 0.00-10.01 sec 2.27 GBytes 1.94 Gbits/sec receiverand this is what I get if the Mac is sending:
Code:
octavarium:~ jollino$ iperf3 -c 192.168.0.50
Connecting to host 192.168.0.50, port 5201
[ 5] local 192.168.0.2 port 50243 connected to 192.168.0.50 port 5201
[ ID] Interval Transfer Bitrate
[ 5] 0.00-1.01 sec 284 MBytes 2.37 Gbits/sec
[ 5] 1.01-2.01 sec 280 MBytes 2.35 Gbits/sec
[ 5] 2.01-3.01 sec 281 MBytes 2.35 Gbits/sec
[ 5] 3.01-4.00 sec 280 MBytes 2.35 Gbits/sec
[ 5] 4.00-5.00 sec 281 MBytes 2.35 Gbits/sec
[ 5] 5.00-6.00 sec 280 MBytes 2.35 Gbits/sec
[ 5] 6.00-7.00 sec 280 MBytes 2.35 Gbits/sec
[ 5] 7.00-8.00 sec 281 MBytes 2.35 Gbits/sec
[ 5] 8.00-9.00 sec 281 MBytes 2.35 Gbits/sec
[ 5] 9.00-10.01 sec 281 MBytes 2.35 Gbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate
[ 5] 0.00-10.01 sec 2.74 GBytes 2.35 Gbits/sec sender
[ 5] 0.00-10.01 sec 2.74 GBytes 2.35 Gbits/sec receiverFrom what I read on a few forums, 5 GbE adapters (based on the Realtek 8157) are not as badly affected but they still show some loss; it's just that at 2.5 Gb there's a lot of headroom before it becomes an issue in that case.
Interestingly, I no longer have Oerrs on the 2.5 GbE interface, whereas I had a LOT of them in Sequoia:
Code:
octavarium:~ jollino$ netstat -di -I en11
Name Mtu Network Address Ipkts Ierrs Opkts Oerrs Coll Drop
en11 1500 <Link#13> 00:e0:4c:68:02:23 16025605 0 2231296 0 0 7
en11 1500 192.168.0 192.168.0.2 16025605 - 2231296 - - -
en11 1500 octavarium. fe80:d::4f0:630c: 16025605 - 2231296 - - -
en11 1500 2a07:7e87:2 2a07:7e87:2865:0: 16025605 - 2231296 - - -
en11 1500 2a07:7e87:2 2a07:7e87:2865:0: 16025605 - 2231296 - - -YMMV but here's to hoping they eventually sort it out, it's not really a dealbreaker but something definitely changed in Sequoia!
For anyone having trouble using AX88179A adapters, I created a small fix using Python. It may not be the best solution for everyone, but it works fine for me: https://github.com/thiago-carneiro/AX88179A_macos_fix