Are there any that actually have a SFP+ port? That's all I want. No one wants to use 10g ethernet when DACs are cheaper than cat7, and you can just change it up to a $7 multimode when you need longer runs.
10G DACs are no cheaper than cat6, which is perfectly fine for 10G at most practical distances. Considering the target audience of these cards it seems pretty obvious to me that letting users "just buy a cat 6 cable" is miles more reasonable than having them buy a transceiver or DAC.
As for allowing to switch to fiber, that just seems orthogonal again to what these USB NICs are for, not to mention the SFP+ itself is probably more expensive than the NIC shown here...
TFA doesn't compare the performance of the new adapters with the older ones.
Does anyone know if the old bulky ones will hit 10G speeds on the same hardware?
I assume I can get a few old TB2 models and adapters on the cheap and they'll run cool enough and stable enough for constant 1G internet and occasional 10G intranet
All these USB version names. I used to know what they all meant, but then the USB IF went ahead and renamed them all and made a bunch of versions have the same name and renamed some versions to have the same name as the old name of other versions.
I have absolutely no idea what anyone means when they say USB 3.2 gen 2x2. I used to know what USB 3.2 meant but it's certainly not that.
To be fair they seem to have taken this often-stated criticism on board. USB 4's naming is more sensible, and they've pushed the simple data speed & power labelling that makes it easier to work out what you need.
The lack of clarity is in keeping with the USB C connector itself, which may supply or accept power at various rates or not at all, may be fast or slow, may provide or accept video or not, and may even provide an interpretation of PCI Express but probably doesn't.
It probably looks the same no matter what, and the cable selected to use probably also won't be very forthcoming with its capabilities either.
It means that if I pick up a random USB cable and plug it into a USB port I have no idea how well it will work or even if it will work at all. It's like the U in USB stands for Unpredictable.
100 mode saved me once when I really really really needed to have a connection in that moment, but the ethernet cable glued to the wall that I was using had only three out of eight wires even functioning.
With 802.3bt type 4 (71W delivered, 90W consumed), absolutely achievable with the proper electronics, but would you trust a no-name, fly-by-night NIC to not fry your expensive devices? That's the biggest hurdle. Possibly a company like Apple, Anker, or similar megacorp or high-trust startup could pull if off.
/*
* RealTek 8129/8139 PCI NIC driver
*
* Supports several extremely cheap PCI 10/100 adapters based on
* the RealTek chipset. Datasheets can be obtained from
* www.realtek.com.tw.
*
* Written by Bill Paul <wpaul@ctr.columbia.edu>
* Electrical Engineering Department
* Columbia University, New York City
/
/
* The RealTek 8139 PCI NIC redefines the meaning of 'low end.' This is
* probably the worst PCI ethernet controller ever made, with the possible
* exception of the FEAST chip made by SMC. The 8139 supports bus-master
* DMA, but it has a terrible interface that nullifies any performance
* gains that bus-master DMA usually offers.
*
* For transmission, the chip offers a series of four TX descriptor
* registers. Each transmit frame must be in a contiguous buffer, aligned
* on a longword (32-bit) boundary. This means we almost always have to
* do mbuf copies in order to transmit a frame, except in the unlikely
* case where a) the packet fits into a single mbuf, and b) the packet
* is 32-bit aligned within the mbuf's data area. The presence of only
* four descriptor registers means that we can never have more than four
* packets queued for transmission at any one time.
*
* Reception is not much better. The driver has to allocate a single large
* buffer area (up to 64K in size) into which the chip will DMA received
* frames. Because we don't know where within this region received packets
* will begin or end, we have no choice but to copy data from the buffer
* area into mbufs in order to pass the packets up to the higher protocol
* levels.
*
* It's impossible given this rotten design to really achieve decent
* performance at 100Mbps, unless you happen to have a 400Mhz PII or
* some equally overmuscled CPU to drive it.
*
* On the bright side, the 8139 does have a built-in PHY, although
* rather than using an MDIO serial interface like most other NICs, the
* PHY registers are directly accessible through the 8139's register
* space. The 8139 supports autonegotiation, as well as a 64-bit multicast
* filter.
*
* The 8129 chip is an older version of the 8139 that uses an external PHY
* chip. The 8129 has a serial MDIO interface for accessing the MII where
* the 8139 lets you directly access the on-board PHY registers. We need
* to select which interface to use depending on the chip type.
*/
For Thunderbolt 4/5 docks, I've held off from buying a high-end Thunderbolt 5 dock as many still have 2.5GbE Ethernet and other limitations with displays. The CalDigit TS5 Plus is one of the only options with 10GbE and its $500 (and usually OoS). I managed to buy an ex-corporate refurb HP Thunderbolt 4 G4 dock for only ~$64 and would recommend others do the same (this has an Intel 2.5GbE and good display outputs)
I have one of these, though I'm using with a USB 3.x port as that's what my desktop has. For me it's working fine, and for others with actual USB 4 ports it seems to be working properly for them.
I am definitely not the person to shed any light on what is going on, but you've added to my feeling that these adapters are all incomprehensible, so I'll try and do the same for you.
I have a USB C ethernet adapter (a Belkin USB-C to Ethernet + Charge Adapter which I recommend if you need it). I ran out of USB C ports one day, and plugged it through a USB C to USB A adapter instead. I must have done an fast.com speed-test to make sure it wasn't going to slow things down drastically, and found that the latency was lower! Not a huge amount, and I think the max speed was quicker without the adapter. But still, lower latency through a $1.50 Essager USB C to USB A adapter, bought from Shein or Shopee or somewhere silly!
I tried tons of times, back and forward, with the adapter a few times, then without the adapter a few times. Even on multiple laptops. As much as I don't want to, I keep seeing lower latency through this cheap adapter.
Next step, I'll try USB C to USB A, then back through a USB A to USB C adapter. Who knows how fast my internet could be!
Too bad this is 10Gbase-T, that energy-wasting hot-running garbage needs to die sooner rather than later. Good thing the ranges for 25Gbase-T are short enough to make it impractical for home use.
(Fibre is nowhere near as "sensitive" as some people believe.)
The problem with fibre isn't the sensitivity. It's that most endpoints have a 1Gbps copper port on them and then Cat6A ports can be used with the common devices but also allow you to add or relocate 10Gbps devices without rewiring the building again.
It's inherently worse than anything fibre, or even DAC cables (which are kinda cheating.) It needs a shitton of analog "magic" to work with the bandwidth limitations of copper cabling.
Will they be cheaper? I look at the RAM prices. Granted,
RAM is in a different category than USB adapters, but
I no longer trust anyone writing "will be cheaper" -
the reality may be different to the projection made.
Are there any that actually have a SFP+ port? That's all I want. No one wants to use 10g ethernet when DACs are cheaper than cat7, and you can just change it up to a $7 multimode when you need longer runs.
10G DACs are no cheaper than cat6, which is perfectly fine for 10G at most practical distances. Considering the target audience of these cards it seems pretty obvious to me that letting users "just buy a cat 6 cable" is miles more reasonable than having them buy a transceiver or DAC.
As for allowing to switch to fiber, that just seems orthogonal again to what these USB NICs are for, not to mention the SFP+ itself is probably more expensive than the NIC shown here...
TFA doesn't compare the performance of the new adapters with the older ones.
Does anyone know if the old bulky ones will hit 10G speeds on the same hardware?
I assume I can get a few old TB2 models and adapters on the cheap and they'll run cool enough and stable enough for constant 1G internet and occasional 10G intranet
All these USB version names. I used to know what they all meant, but then the USB IF went ahead and renamed them all and made a bunch of versions have the same name and renamed some versions to have the same name as the old name of other versions.
I have absolutely no idea what anyone means when they say USB 3.2 gen 2x2. I used to know what USB 3.2 meant but it's certainly not that.
To be fair they seem to have taken this often-stated criticism on board. USB 4's naming is more sensible, and they've pushed the simple data speed & power labelling that makes it easier to work out what you need.
I don't think they've taken the criticism on board, USB 3 still has the completely nonsensical names
Oh, it's fine.
The lack of clarity is in keeping with the USB C connector itself, which may supply or accept power at various rates or not at all, may be fast or slow, may provide or accept video or not, and may even provide an interpretation of PCI Express but probably doesn't.
It probably looks the same no matter what, and the cable selected to use probably also won't be very forthcoming with its capabilities either.
(Be sure to drink your Ovaltine.)
What difference does that make in your life?
This article we are discussing this about gives a great example of why understanding the difference matters a lot when purchasing hardware.
It means that if I pick up a random USB cable and plug it into a USB port I have no idea how well it will work or even if it will work at all. It's like the U in USB stands for Unpredictable.
2 replies →
A sense of perfect knowledge of the things around you, and not feeling like someone is trying to take advantage of you any time you go shopping.
Well it means I have no idea what TFA is talking about when it mentions USB versions, for one.
2 replies →
A Framework expansion card was also announced this week. https://frame.work/nl/en/products/wisdpi-10g-ethernet-expans...
That link notes:
"Card supports 10Gbit/s and 10/100/1000/2500/5000/10000Mbit/s Ethernet"
Nice to see; some NICs are shedding 10/100 support. Apparently, it's not necessary to do this, even in a low cost device.
100 mode saved me once when I really really really needed to have a connection in that moment, but the ethernet cable glued to the wall that I was using had only three out of eight wires even functioning.
1 reply →
100 is needed for embedded stuff, it'd render a lot of devices unusable (wiznet chips are popular and are 100 only). That'd suck.
Low-cost devices are exactly where 10/100 is still widely used. On PCs, it's a common power-saving mode.
4 replies →
Lots of industrial sensors and devices only do 4 wire 100BASE-TX so if there's no fallback to that it would be a paperweight in those situations.
-
7 replies →
Is it also possible to power a laptop through those adapters? PoE++ can deliver up to 100W of power, more than enough for most laptops.
Theoretically yes, practically that hasn't been built yet. I've only seen it for 2.5Gbase-T, and only for 802.3bt Type 3 (51W).
If anyone's aware of something better, I'd be interested too :)
(Then again I wouldn't voluntarily use 5Gb-T or 10Gb-T anyway, and ≈50W is enough for most use cases.)
[ed.: https://www.aliexpress.us/item/3256807960919319.html ("2.5GPD2CBT-20V" variant) - actually 2.5G not 1G as I wrote initially]
Eh.
A lot of laptops won't accept less than 60w
My work laptop won't accept less than 90w (A modern HP, i7 155h with a random low end GPU)
At first everyone at the office just assumed that the USB C wasn't able to charge the pc
18 replies →
The idea of a POE Mac mini makes me happy. It would be a nice way of power cycling it from the switch, tidier than the smart plug I have.
https://hackaday.com/2023/08/14/adding-power-over-ethernet-s...
We used PoE hats for a bunch of Raspberry Pis once. It’s definitely a great idea.
I can’t find what you want, but you can buy PoE splitters. PoE in, ethernet and power out.
Surely a matter of time until someone does this…
I found a 5gbe one that claimed 60W, will power a phone but not the low power laptop I've got here. It probably isn't far off.
I think class 4 tops out at about 71W delivered to the powered device, albeit 90W at the switch port.
Might be a struggle I suspect!
With 802.3bt type 4 (71W delivered, 90W consumed), absolutely achievable with the proper electronics, but would you trust a no-name, fly-by-night NIC to not fry your expensive devices? That's the biggest hurdle. Possibly a company like Apple, Anker, or similar megacorp or high-trust startup could pull if off.
Yes, but look up the prices for PoE switches and you might reconsider.
By the way, how are switches and cables for > 1Gbps these days?
From the source of the RealTek 8129/8139 PCI NIC driver in FreeBSD: https://elixir.bootlin.com/freebsd/v10.2/source/sys/pci/if_r...
/* * RealTek 8129/8139 PCI NIC driver * * Supports several extremely cheap PCI 10/100 adapters based on * the RealTek chipset. Datasheets can be obtained from * www.realtek.com.tw. * * Written by Bill Paul <wpaul@ctr.columbia.edu> * Electrical Engineering Department * Columbia University, New York City / / * The RealTek 8139 PCI NIC redefines the meaning of 'low end.' This is * probably the worst PCI ethernet controller ever made, with the possible * exception of the FEAST chip made by SMC. The 8139 supports bus-master * DMA, but it has a terrible interface that nullifies any performance * gains that bus-master DMA usually offers. * * For transmission, the chip offers a series of four TX descriptor * registers. Each transmit frame must be in a contiguous buffer, aligned * on a longword (32-bit) boundary. This means we almost always have to * do mbuf copies in order to transmit a frame, except in the unlikely * case where a) the packet fits into a single mbuf, and b) the packet * is 32-bit aligned within the mbuf's data area. The presence of only * four descriptor registers means that we can never have more than four * packets queued for transmission at any one time. * * Reception is not much better. The driver has to allocate a single large * buffer area (up to 64K in size) into which the chip will DMA received * frames. Because we don't know where within this region received packets * will begin or end, we have no choice but to copy data from the buffer * area into mbufs in order to pass the packets up to the higher protocol * levels. * * It's impossible given this rotten design to really achieve decent * performance at 100Mbps, unless you happen to have a 400Mhz PII or * some equally overmuscled CPU to drive it. * * On the bright side, the 8139 does have a built-in PHY, although * rather than using an MDIO serial interface like most other NICs, the * PHY registers are directly accessible through the 8139's register * space. The 8139 supports autonegotiation, as well as a 64-bit multicast * filter. * * The 8129 chip is an older version of the 8139 that uses an external PHY * chip. The 8129 has a serial MDIO interface for accessing the MII where * the 8139 lets you directly access the on-board PHY registers. We need * to select which interface to use depending on the chip type. */
8159 != 8139
> /* * RealTek 8129/8139 PCI NIC driver * * Supports several extremely cheap PCI 10/100 adapters based on […]
Also, please, for the love of whatever entity, at least remove the *s on that paste. This is just atrocious and disrespectful of any reader.
My favorite USB ethernet adapter is a lowly 100 MBit one that works everywhere without requiring driver downloads.
FWIW I got a Xikestor 10G adapter with the Realtek chipset from AliExpress and it underperforms my much cheaper 5G one.
Yeah. Just because it negotiates, doesn’t mean it can utilise.
For Thunderbolt 4/5 docks, I've held off from buying a high-end Thunderbolt 5 dock as many still have 2.5GbE Ethernet and other limitations with displays. The CalDigit TS5 Plus is one of the only options with 10GbE and its $500 (and usually OoS). I managed to buy an ex-corporate refurb HP Thunderbolt 4 G4 dock for only ~$64 and would recommend others do the same (this has an Intel 2.5GbE and good display outputs)
The PCIe version: https://news.ycombinator.com/item?id=46423967
It seems like a lot of laptop manufacturers skipped the USB 3.2 Gen2x2 in favor of USB4/TB4.
Conversely, the last time I checked a couple of weeks ago, it was impossible to find any USB4 external SSDs on Amazon; only USB 3.2.
Wouldn't it be better to just buy an M.2 NVMe adapter, eg. ICY DOCK ICYNano MB861U31-1M2B[0]?
[0]: https://global.icydock.com/product_247.html
3 replies →
If Amazon is a strict requirement, then this won't help. But if you're ok with AliExpress then it's probably a win:
https://www.aliexpress.com/item/1005008555989592.html
I have one of these, though I'm using with a USB 3.x port as that's what my desktop has. For me it's working fine, and for others with actual USB 4 ports it seems to be working properly for them.
Really? I see plenty when I search for 'usb4 nvme enclosure'
I have a RTL8157 5 Gbps adapter from CableMatters.
Interestingly it seems to get burning hot on the MacBook M1 Pro while it remains cool on the M5 Pro model.
Maybe the workload is different, but I would not rule out some sort of hardware or driver difference. I only use a 1G port on my router at the moment.
Huh! That's very interesting.
I am definitely not the person to shed any light on what is going on, but you've added to my feeling that these adapters are all incomprehensible, so I'll try and do the same for you.
I have a USB C ethernet adapter (a Belkin USB-C to Ethernet + Charge Adapter which I recommend if you need it). I ran out of USB C ports one day, and plugged it through a USB C to USB A adapter instead. I must have done an fast.com speed-test to make sure it wasn't going to slow things down drastically, and found that the latency was lower! Not a huge amount, and I think the max speed was quicker without the adapter. But still, lower latency through a $1.50 Essager USB C to USB A adapter, bought from Shein or Shopee or somewhere silly!
I tried tons of times, back and forward, with the adapter a few times, then without the adapter a few times. Even on multiple laptops. As much as I don't want to, I keep seeing lower latency through this cheap adapter.
Next step, I'll try USB C to USB A, then back through a USB A to USB C adapter. Who knows how fast my internet could be!
[dead]
[dead]
Too bad this is 10Gbase-T, that energy-wasting hot-running garbage needs to die sooner rather than later. Good thing the ranges for 25Gbase-T are short enough to make it impractical for home use.
(Fibre is nowhere near as "sensitive" as some people believe.)
The problem with fibre isn't the sensitivity. It's that most endpoints have a 1Gbps copper port on them and then Cat6A ports can be used with the common devices but also allow you to add or relocate 10Gbps devices without rewiring the building again.
However — unlike copper twisted pair — the bandwidth current fiber media can carry is nearly limited by nothing but the optics at each end.
2 replies →
In practice though 10G via copper requires pretty perfect terminations. The slightest error leads to crosstalk issues.
1 reply →
Good thing the ranges for 25Gbase-T are short enough to make it impractical for home use.
Anyone who talks about 25GBASE-T like it actually exists, doesn't know anything about what they're talking about.
Or is speaking in future terms.
40Gbase-T will never exist, sure. 25Gbase-T very likely will.
Is the energy consumption inherent to 10Gbase-T? Or is it that 1Gbit nics have been around forever and optimised ad infinitum?
To be fair, the power consumption is also my biggest gripe with my WiFi 6 AP, they run extremely hot.
It's inherently worse than anything fibre, or even DAC cables (which are kinda cheating.) It needs a shitton of analog "magic" to work with the bandwidth limitations of copper cabling.
Will they be cheaper? I look at the RAM prices. Granted, RAM is in a different category than USB adapters, but I no longer trust anyone writing "will be cheaper" - the reality may be different to the projection made.
Can these support local LLM’s?