Review Request: WisdPi USB 3.2 5Gbps Wired Base-T Ethernet Adapter (WP-UT5) based on Realtek RTL8157

[talynone]

There's a three year old review roundup of 5gbe to USB adapter all basically based on the Marvel chipset:

Jeff Geerling recently did a review of the new Realtek based 5gbe to USB adapter worth a look, the WisdPi
USB 3.2 5Gbps Wired Base-T Ethernet Adapter (WP-UT5) Realtek RTL8157.

It'd be nice to have a more comprehensive STH review of it.

Can be purchased here:

Realtek RTL8157 USB 3.2 5Gbps Wired Base-T Ethernet Adapter (WP-UT5)

WP-UT5 is the first batch is expected to be shipped before August 1st 2024. The first 50 orders can use the coupon(B5H8BWKS7VTT) to enjoy a $5 discount. Thank you for your support. Features Chip: Realtek RTL8157. All support 10/10/100/2500/5000 Mbps links and consumes under 1.5 watts, against...

www.wisdpi.com

 

[MountainBofh]

Can be found on Amazon for under $35

https://www.amazon.com/Ethernet-Adapter-WP-UT5-Realtek-RTL8157/dp/B0DF2TTT9H

 

[jerrythea]

There is also a Wavlink version using the same chipset.

A more main-stream brand, but not necessarily good quality.

Amazon reviews are quite mixed, but it looks like they mix up all of the various versions of this USB LAN (some gigagbit, some 2.5, etc).

Also, the reviews are not necessarily from informed readers.

https://www.amazon.com/WAVLINK-Ethernet-5000Mbps-Converter-Aluminum/dp/B0DBZ41BPL/ref=cm_cr_arp_d_bdcrb_top?ie=UTF8&th=1

 

[talynone]

Looks like the driver is required to get full speed which seems to max out at around 4.7 gigabits. Much better than the Marvel chipset. One of the Amazon reviews of the WAVLINK has a video of a speedtest going 4.7 gigabits in OSX. Jeff has a test of the WisdPi in Windows 11 of it going that speed once the drivers were installed.

The WisdPi product page indicates the Realtek Windows 10 driver is unstable, Windows 11 should be fine.

The WisdPi wiki has instructions on how to load the driver for Linux, Proxmox and Synlogy NAS systems:

USB 3.2 5Gbps Wired Base-T Ethernet Adapter | WisdPi

Model: WP-UT5

wiki.wisdpi.com

 

[MountainBofh]

I wonder if their driver is the newest or not. Newest drivers can be found here.

Realtek

www.realtek.com

 

[jerrythea]

So the next question is which smart managed switch supports multi-gigabit?

the best option (ignoring price) appears to be the Netgear:

Smart Switches - MS510TXUP | NETGEAR

Looking for Smart Switches? Discover Netgear MS510TXUP 8-Port Multi-Gigabit/10G Ethernet Ultra60 PoE++ Smart Switch with 2 SFP+ Ports. Buy Now!

www.netgear.com

 

[jerrythea]

Or perhaps this one, without PoE++

Smart Switches - MS510TXM | NETGEAR

Looking for Smart Switches? Discover Netgear MS510TXM 8-Port Multi-Gigabit/10G Ethernet Smart Switch with 2 SFP+ Ports. Buy Now!

www.netgear.com

 

[MountainBofh]

The Unifi XG24 supports it. I know a lot of the newer enterprise switches do as well.

 

[talynone]

Someone just did a video of the Wavlink model in action:

 

[Zerreth]

I received the Wavlink WL-NWU340G-A today which uses the same realtek chipset, and it's currently super unstable with the latest Realtek drivers.

Before driver installed (windows auto assigns driver microsoft driver from 2016):


After the driver install you can get 5Gbps, BUT....
Whenever you perform a full load test on it, USB drops out.

Rear USB-C:


Front USB-C: (other devices are plugged in as well so it's probably sharing bandwidth)


This is on an AMD B550M platform, with latest chipset drivers installed. USB drop-outs happen on both front & rear USB-C connections.

 

[talynone]

I received the Wavlink WL-NWU340G-A today which uses the same realtek chipset, and it's currently super unstable with the latest Realtek drivers.

Are you running WIndows 10 or Windows 11? The driver download page says the drivers are unstable on Windows 10, but should be ok on Windows 11.

 

[Zerreth]

Windows 11 Pro Version 10.0.22631 Build 22631

 

[Zerreth]

Re-tested the Wavlink WL-NWU340G-A on a HP Spectre 15 (Intel i7-8750H) & Windows 11 Home Version 10.0.22631 Build 22631
Performance is heavily dependant on the power state.

One weird quirk with it is that I always have to unplug & replug after reboot, otherwise webpages endlessly load.

On Battery (3.2Gbps):

On 80W USB Power Delivery (3.6 - 4.1Gbps):

With Included Barrel Charger(4.5 - 4.8Gbps):

 

[MountainBofh]

I ordered the Wavelink. I'll test it on a HP Spectre X360 laptop with a 10th gen I7 running Windows 10 LTSC 21H2.

Curious to see how unstable it will be when using the latest drivers from Realtek.

 

[LeoRX]

I've ordered and returned the Wavlink unit as it's not stable in Unraid. NIC should just work,.. I was getting drop out once a day or once every couple of hours.
Has anyone purchased and used WisdPi unit?

 

[jerrythea]

It's bleeding edge.

With Proxmox or other Linux Kernel on 6.11++, and the driver provided on the realtek site, I have had good luck on almost all requirements.

iperf3, scp, etc get expected performance.

However, I am now having an issue with NFS server on top of this NIC. It is odd, but it locks up the CPUs, sometimes in unrelated network drivers (igc-intel 226v) and sometimes in r8152 (custom driver).

If you need it to just work, you will need either a PCIe slot, or get the thunderbolt 10GbE NIC from OWC. It runs really hot!, and is not cheap (~$150) but it might be more stable.

 

[tusk9541]

I've been using the Wavlink on a Thinkpad P16s AMD 7840u. Out of the USB4 port I have a Kensington SD2600T Thunderbolt 4 hub. Connected directly to one of the TB downstream ports, the Wavlink maxes out on iperf3, ~4.74Gb/s, both upload and download. Connected to the USB-A 10Gbps port in the front with an adapter, I get full upload, but the download drops to ~2-3 Gb/s.

But... I've also experienced disconnections. Before taking the hub out of the equation I decided to search and it appears that this thing is still unstable with the latest Realtek drivers 1157.15.20.327, so I'm gonna wait cause testing without the hub is going to be a bit of a hassle due to my setup and lack of space.

Also I've noticed in the Amazon listing they've added another "style", besides the "340G" that we have, there's a "341G" now, without any discernible difference, with identical photos and description. One wonders if the disconnections aren't due to the Realtek chip and something else in the 340G is faulty. I'm past my return date, but if anyone that has the 340G and can still return it wants to test...

 

[jerrythea]

I setup the thunderbolt 10 GbE from OWC and have had no issues (except for a required udev rule to ensure auto activation for Thunderbolt devices).

So I now have a home built NAS based on ASRock NUCBox (Intel Ultra 5) running 10 GbE against 3 clients using the Wisdpi 5 GbE adapters.

I have settled on just using Fedora server for the NAS given the confusing NAS options (I will never go back to Synology).

All of the VMs (Proxmox 8.2 ) have their storage on the NAS via a dedicated MicroTik 10 GbE switch) using the WisdPi’s 5 GbE for each client. This seems like a good justification for the cost of the OWC 10 GbE adapter.

The cluster uses a separate dedicated 2.5 GbE Switch for inter cluster communications.

I have another NAS that I want to also use on the higher speed network, but am wary of using the WisdPi with NFS, based on the headaches).

I had experimented with shared LVM over iSCSI but managed to blow away the whole LVM volume multiple times (maybe user error).

NFS is slower, but more resilient and easier to manage.

 

[jerrythea]

Wisdpi to WisdPi via Microtik switch:


Some sample file transfer numbers:

# scp -pr *.iso 192.168.90.4:/mnt/pve/pve4vm/tm
ClearOS-DVD-x86_64.iso                                                       100% 1224MB 541.0MB/s   00:02
debian-10.6.0-amd64-netinst.iso                                              100%  349MB 558.3MB/s   00:00
debian-11.6.0-amd64-netinst.iso                                              100%  388MB 542.8MB/s   00:00
debian-12.6.0-amd64-netinst.iso                                              100%  631MB 549.8MB/s   00:01
debian-9.13.0-amd64-DVD-1.iso                                                100% 3574MB 555.2MB/s   00:06
debian-9.13.0-amd64-DVD-2.iso                                                100% 4458MB 550.9MB/s   00:08
debian-9.13.0-amd64-DVD-3.iso                                                100%  423MB 550.5MB/s   00:00
foo.iso                                                                       81% 8835MB 550.8MB/s   00:03 ETA

 

[John T Davis]

Just testing one of these today, using Proxmox.
At least on PVE's stable kernel (6.8.12-10-pve), the necessary Realtek drivers aren't installed.

Without the drivers, iperf3 couldn't break 2 Mbps and was flooding me with thousands of re-transmits.
I also appreciate that it pins the name of the USB adapter so Debian can't change it.

I installed the DKMS version of the driver, here: GitHub - wisdpi/wp-ut5_dkms: A DKMS package for easy use of WP-UT5
I went with the DKMS version because Proxox updates the kernel so frequently, and this should auto-build the module every time there's an update.

It seems to have worked:

[ 2381.106799] r8152 3-1:1.0: ram code version is wrong 0000(0013
[ 2381.205781] r8152 3-1:1.0 eth0: v2.19.2 (2024/10/21)

Everything looks good:

~/build-files/wp-ut5_dkms# ethtool eth0
Settings for eth0:
        Supported ports: [ MII ]
        Supported link modes:   10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
                                1000baseT/Full
                                2500baseT/Full
                                5000baseT/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
                                2500baseT/Full
                                5000baseT/Full
        Advertised pause frame use: No
        Advertised auto-negotiation: Yes
        Advertised FEC modes: Not reported
        Link partner advertised link modes:  10baseT/Full
                                             100baseT/Full
                                             1000baseT/Full
                                             2500baseT/Full
                                             5000baseT/Full
        Link partner advertised pause frame use: No
        Link partner advertised auto-negotiation: Yes
        Link partner advertised FEC modes: Not reported
        Speed: 5000Mb/s
        Duplex: Full
        Auto-negotiation: on
        Port: MII
        PHYAD: 32
        Transceiver: internal
        Supports Wake-on: pumbg
        Wake-on: d
        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

Now I've got an interesting problem. I'm not actually sure if it's a problem, or just the nature of multigig.

$] iperf3 -B 10.10.200.19 -c 10.10.200.12 -P 5 -R

[SUM]   0.00-10.00  sec  5.65 GBytes  4.85 Gbits/sec  13643             sender
[SUM]   0.00-10.00  sec  5.65 GBytes  4.85 Gbits/sec                  receiver

(Using 20 threads pushes it to 4.9 Gbps.)

That is a lot of retries. Is that actually a problem? I know that in -R mode, the server sends and the client recieves. In this case, the server has a full 10 Gbps connection (2x10Gbps LACP), and the test client only has the 5 Gbps connection.

I'm hoping the server is just flooding it too fast, which isn't actually a problem … right?

Zero (0) retries with: $] iperf3 -B 10.10.200.19 -c 10.10.200.12 -P 5 .