Live data from Hacker News

New 10 GbE USB adapters are cooler, smaller, cheaper

jeffgeerling.com

191–200 of 384 posts

Re: New 10 GbE USB adapters are cooler, smaller, cheaper

#191
post #9

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.

Doing home automation of lamps, sensors, speakers via PoE would be great too. It should faster and more stable than Zigbee/Wifi and with no need to change the batteries often.

Re: New 10 GbE USB adapters are cooler, smaller, cheaper

#192
post #76
post #64

Earlier quoted context omitted.

[flagged]

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.

Get a certified cable if it matters to you. If you pick a random cable out of the cable box it’s probably a 2.0 charging only cable that hasn’t been certified by USB IF.

Ultimately the majority of people only use usb cables for charging or 2.0 speeds for their keyboard or mic so this isn’t a problem. And for those who it is a problem, they know which one their high speed cable is.

Re: New 10 GbE USB adapters are cooler, smaller, cheaper

#193
post #63

Earlier quoted context omitted.

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 eithe…

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 gets even worse. I now have two cheap Chinese gadgets (a checki printer and a tire inflater) that have USB-C ports for charging, but will only charge wi…

I keep a few of these around to deal with this: https://www.adafruit.com/product/6323

Very annoying though! The devices are just missing a couple resistors which is probably less than a cent on the BOM.

Re: New 10 GbE USB adapters are cooler, smaller, cheaper

#194

Earlier quoted context omitted.

10 Gb is cheap! Mikrotik has a 4x10Gb + 1x1Gb port switch for $150 USD and an 8x10Gb version for about $275. I have the 8 port version. SFP+'s and fiber are cheap, like maybe 50 bucks for the SFP+ set and fiber. 10Gb PCIe cards are maybe ~$50 new on Amazon with Intel chips and cheaper on eBay - I bought used 10 Gb Mellanox cards for $25 each - "they just work" under FreeBSD and Linux. Copper 10 Gb used to consume waa…

10 Gb is cheap! … $150 … $275. San Francisco checking in.

A single eero or Ubiquiti AP will be $150-300 depending on the exact capabilities, so if you're pricing out how to network your house I'd say the switch looks pretty good b

Re: New 10 GbE USB adapters are cooler, smaller, cheaper

#195
post #84
post #5

Earlier quoted context omitted.

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 is needed for embedded stuff, it'd render a lot of devices unusable (wiznet chips are popular and are 100 only). That'd suck.

IKEA smart home hub is also 100mbit.

Re: New 10 GbE USB adapters are cooler, smaller, cheaper

#198

Earlier quoted context omitted.

There’s nothing hard about it if you can run pre-terminated patches. Which you typically can since the connectors are so small.

So you're saying users could buy stuff like this? "25m (82ft) Fiber Patch Cable, 1 Fiber, SC APC Simplex to SC APC Simplex, Single Mode (OS2), Riser (OFNR), 2.0mm, Tight-Buffered, Yellow", https://www.fs.com/eu-en/products/282133.html?attribute=1031... Heck, I don't even know what I should buy for 10G SFP+ ports and a distance of say 30 meters. Guess, I'm back to CAT6 :-)

> Guess, I'm back to CAT6 :-)

If you learned what you need for 10GbT you can learn what you need for 10GbLR. Which is:

LC connector, PC or UPC, duplex, OS1 or OS2, and SFP+ modules saying "LR".

Any of the following is wrong: SC, FC, LSH, E2000, ST, APC, simplex, OM[1-5], "SR" or "ER" SFPs.

And that's short enough.

Re: New 10 GbE USB adapters are cooler, smaller, cheaper

#199

Earlier quoted context omitted.

The fact that you had to list all of the versions and speeds at the top of your post is illustrative of what the parent was trying to say. We can all look up what speed is associated with what version, but it’s not exactly a consumer friendly experience.

A few computer manufacturers do the right thing and they mark the speed on the USB ports, removing ambiguities, for example ASUS does this on my NUCs and motherboards. Unfortunately, there are too many who do not do this, even among the biggest computer vendors.

> mark the speed on the USB ports, removing ambiguities

Unfortunately it's not true.

Quiz: what happens when a device capable of 20Gbps is plugged into a port marked as 40Gbps?

Re: New 10 GbE USB adapters are cooler, smaller, cheaper

#200
I'm disappointed that both the article and comments don't go into the actual differences between how these adapters work and the overhead incurred by USB.

At a high level, I'm pretty sure Thunderbolt will be significantly better in all situations:

Thunderbolt is PCIe; depending on the way the network card driver works, the PCIe controller will usually end up doing DMA straight into the buffers the SKB points to, and with io_uring or AF_XDP, these buffers can even be sent down into user space without ever being copied. Also, usually these drivers can take advantage of multiple txqueues and rxqueues (for example, per core or per stream) since they can allocate whatever memory they want for the NIC to write into.

USB is USB; the controller can DMA USB packet data into URBs but they need to be set up for each transaction, and once the data arrives, it's encapsulated in NCM or some other USB format and the kernel usually has to copy or move the frames to get SKBs. The whole thing is sort of fundamentally pull based rather than push based.

But, this is just scratching the surface; I'm sure there are neat tricks that some USB 3.2 NIC drivers can do to reduce overhead and I'd love to read an article where I learned more about that, or even saw some benchmarks that analyzed especially memory controller utilization, kernel CPU time, and performance counters (like cache utilization). Especially at 10G and beyond, a lot of processing becomes memory bandwidth limited and the difference can be extremely significant.

Post reply on HN