The RTL8157 chipset works with 2.18-1-1. #428
Comments
|
WP-UT5 is now available for purchase on : |
|
Same here, the MTU cannot be set either via Synology tool or the shell |
|
This is lack of Link Speed is also an issue for Active/Standby on bonding. The below |
@yunhao-jiang Many 10Gbps SFP modules do not negotiate the 5Gbps speed correctly, although they may function normally. You need to use an SFP module that supports the 5Gbps speed for testing. We will investigate this issue. |
|
@yunhao-jiang Tested works. Speeds exceeding 5000Mb/s and bond supported. |
|
I have the same problem =============================== I actually confirmed that data transfer at 5GbE speed is possible. The product used is WisdPi WT-UT5. RTL8152/RTL8153 driver [2.18-1-1] ========================================== |
|
I am also having issues with WP-UT5 RTL8157 connected to a DS1513+ running DSM 7.1.1-42962 Update 6 running driver 2.18.1-1. My main issue is that no matter what I try on the Synology, speeds are always capped at 1Gbps max even when I force the speed to 2.5 or 5. Test environment: Client side used to test speeds: Windows 11 with Intel X540-AT2 connected at 10Gbps Tested on both USB 3 ports on the Synology; hardware reports correct USB speed required: 4901.972688] usb 8-1: reset SuperSpeed USB device number 2 using etxhci_hcd-170202 Each Tranciever was tested with the same results below. Note that on the switch, the port shown as expected on the SFP+ side ad connecting at 10Gbps and the RJ-45 side speed is not able to be forced to a specific speed. sudo ethtool eth4 | grep Speed sudo ethtool eth4
and you even get it to try to act like a team player and state its link speed; buts its not true. but there really is no actual change. I assume that the driver needs a bit more work to fully support this NIC. @bb-qq and @wisdpi may have a few comments of what to try next. Happy times on the bleeding edge train flying close to the sun. |
|
@rsockanc The speed you are seeing does indeed seem abnormal. We have identified the reason why Synology cannot negotiate speeds exceeding 2500Mbps and are working on resolving it. However, we may release a WisdPi version of the driver, which is still under testing. |
|
Glad to hear that @wisdpi . One possibly related issue I just tested: When the WisdPi WT-UT5 is connected to a Windows 11 PC with the USB-C to USB-C cable, link speeds negotiate and transmit/receive correctly as expected at 1, 2.5, and 5Gbps. However, when connected with the USB- A to USB- C supplied cable, the link negotiates correctly to 5, 2.5, or 1 but is only able to transmit reliably when set to 1 or 2.5 Gbps. When set to 5 Gbps with the with the USB- A to USB- C, the nic will ramp up a download up to 2.1Gbps for 3-8 seconds then will start slowing down until it fails to transmit any more packets without a reset. |
|
Same here, seems like 5Gbps is not quite enough for this adapter? The USBC is connected through 10Gbps USBC channel while USB-A is at most 5Gbps I think. I have better experience when I use a thunderbolt cable for some reason.
|
|
I noticed a few things and will leave a comment.
|
Looks like the reason why DS920 won't show 5Gbps is the kernel version too low. On my end the most up-to-date system shows 4.4.302, which is still lower than 4.10 where 5G link negotiation is supported. |
@rsockanc @yunhao-jiang USB A-C can also provide a speed of 10Gbps, so it requires the USB 3.2 standard. The instability issue we suspect is due to many USB A ports not being able to provide sufficient power. All cables are tested for 10Gbps speed when they leave the factory. Both A-C and C-C cables undergo speed testing. |
That's right. The Realtek driver need kernel 4.10 +. Only SA6400 is fine. |
|
The WP-UT5 uses a 10Gbps USB flipping chip, and the effect is the same whether inserted forwards or backwards. We have tested each data cable, but not for an extended period. There is indeed a quality risk with this. |
That's a bit confusing. What am I missing? |
@bb-qq @RieGo I just wrote a tutorial on using the WP-UT5 on Ubuntu and tested this specific command. I believe there is an error in the readme provided by Realtek. After compiling the Linux driver, the WP-UT5 works perfectly and negotiates correctly. However, when I used the command (ethtool -s eth0 autoneg on advertise 0x180000000002f) to set the speed to 5Gbps, the network card speed changed from 5Gbps to 2.5Gbps. |
|
Do we have a workaround on the mtu problem yet? It seems this is a bug in the Realtek driver. |
@RieGo @bb-qq @i0ntempest @rsockanc @nogcha234 We have optimized the Realtek Linux driver, which now supports up to a 16K MTU. This has been synchronized with our Linux driver codebase (https://github.com/wisdpi/wp-ut5_linux), and we will soon apply to merge the code into the bb-qq code repository.
|
|
Thank you @wisdpi. I have not confirmed that it works on my end, but I think the risk of a fix is low, so I have released a package with the above patch applied. |
That's fine. I will check DSM later. I'm sure that it will be work. |
|
I updated driver to 2.18.1-2 on my DS 718+ (newest DSM 7.2.2-72806). The DSM GUI still does not show option to select MTU higher than 1500. UPDATE: MTU can be set to jumbo frame via command line (verify device name, and substitute as necessary):
|
|
Just to confirm: The only Synology model that can use the RTL8157 at 5Gbe is the SA6400. Because all the rest have Linux Kernel below 4.10, and the max speed would be 2.5Gbe. Is that correct? |
|
I am seeing about 3 Gbps NAS-to-NAS (both DS718+) through 10 Gbps switch using 9K MTU running test with iperf3. I am guessing that USB 3.0 ports may be limitation. |
@joshuacant WisdPi RTL8157 WP-UT5 test result on DS1819+（Kernel 4.4） with USB 5Gbps. About 3.15Gbps MTU 1500. DS1819+ only have USB 3.0 5Gbps.
|
|
Thank you for your replies! I must be doing something wrong. I have a 920 and here's what I see after first plugging the device in.
The speeds are limited to 1Gbe and the "network status" is missing from the Synology Network GUI. If I run the command
I know my network cables and switch are capable of 5Gbe because I have been using the aqc111 device for many months. |
|
|
@wisdpi @bb-qq I initially had version 2.18-1-1 of the package installed. After a reboot, I had a network connection via the front USB (back didn't seem to work). I did not have any options like Jumbo Frames enabled, and the web interface didn't properly display the negotiation speed. When I updated the driver to version 2.18-1-2, the speed suddenly dropped considerably (~40MB/s - USB2.0??). After downgrading, I'm back at ~370MB/s. By the way, I also have the 920+, so I'm on an older kernel version. My findings seem to align with the experiences of @nogcha234 and @joshuacant. |
|
Hi, I am wondering if there has been any progress toward getting the 8157 chipset working with both large MTU and broadcasting 5GbE link mode with older kernels, or is this even possible? Thank you. |
|
From what I can tell from other posts, we may not be fighting just the kernel version, but also the hardware capabilities of the USBPHY that was architected for the older devices back in the day. I appreciate that we are all trying to continue to use storage hardware that is over 10 years old, but at some point, we might have to go purchase something new. |
|
Well, I had been using bb-qq's excellent aqc111 driver, so I know what speeds should be the maximum possible with the RTL8157 on older hardware. I could back to using the aqc111 device but they run quite hot and can be unstable. So I am hopeful this driver will get there. |
|
I'm using a DS920+ with version 2.18-1-2. I can achieve 2.71 Gbps (my TP-Link TL-SX105 shows it as a 5 Gbps link) when letting the adapter decide the speed (though it shows the speed as "unknown") and 2.38 Gbps (my TL-SX105 shows it as a 2.5 Gbps link) when forcing the adapter to 2.5 Gbps. I noticed some Amazon reviews mentioning a 430 MB/s speed on the DS920+ with a 9K MTU setting. Not sure how that user achieved it. @RieGo Regarding USB 2.0 speeds, have you tried flipping the USB-C connector? I had the same issue and noticed others discussing it, which resolved it for me by flipping the connector. |
The RTL8157 chipset works with 2.18-1-1.
2.18-1.1 package shows the RTL8157 chipset in action.
I actually confirmed that data transfer at 5GbE speed is possible.
However, I can't check the link speed connected to 5000Mbps (5GbE) in Synology's ethtool or Synology network information, and I can't modify the MTU.
This problem is unlikely to be fixed right away, but I'll leave it here for reference.
The product used is WisdPi WT-UT5.
|__2-2 0bda:8157:3000 00 3.20 5000MBit/s 544mA 1IF (WisdPi USB 5G Ethernet 000334C******10017)
eth2 Link encap:Ethernet HWaddr 34:::B1::17
inet addr:10.1.0.* Bcast:********* Mask:255.255.255.0
inet6 addr: fd3b::bee2::36c8:d6ff:feb1:17/64 Scope:Global
inet6 addr: fe80:::d6ff:*:17/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:141772521 errors:0 dropped:0 overruns:0 frame:0
TX packets:390506389 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:198520946899 (184.8 GiB) TX bytes:552978638476 (515.0 GiB)
Settings for eth2:
Supported ports: [ MII ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
2500baseX/Full
Supported pause frame use: No
Supports auto-negotiation: Yes
Supported FEC modes: Not reported
Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
2500baseX/Full
Advertised pause frame use: Symmetric Receive-only
Advertised auto-negotiation: Yes
Advertised FEC modes: Not reported
Link partner advertised link modes: 100baseT/Half 100baseT/Full
1000baseT/Half 1000baseT/Full
Link partner advertised pause frame use: No
Link partner advertised auto-negotiation: Yes
Link partner advertised FEC modes: Not reported
Speed: Unknown!
Duplex: Full
Port: MII
PHYAD: 32
Transceiver: internal
Auto-negotiation: on
Cannot get wake-on-lan settings: Operation not permitted
Current message level: 0x00007fff (32767)
drv probe link timer ifdown ifup rx_err tx_err tx_queued intr tx_done rx_status pktdata hw wol
Link detected: yes
The text was updated successfully, but these errors were encountered: