Intel Sends Out Initial USB 4.0 Support For The Linux Kernel

[Teggs]

I have been wondering for the last few years: Why haven't the people who keep messing this up been removed?

No users seem to like confusion in standards and interoperability.

No users seem to like confusion in (re)naming.

Yet this goes on and on. I don't understand why major players don't just sit on or replace the USB-IF, change all of its directors by force, whatever it would take to end this problem.

Logic suggests someone is profiting from the situation, but I don't understand who or where. It isn't like a user can easily replace the USB on their device with a superior one (or one that does what they want), so how does the fragmentation generate profit?

[anth]

there's only really 2 types of USB-C cable, the thin/cheap 4-wire type that only supports USB2.0 and the full featured type that supports the rest

With USB 3.1 and 3.2 there were 6 types of cable with USB-C at each end. Three speeds (USB 2.0, 5Gb/s per lane, 10 Gb/s per lane) multiplied by two current ratings (3A, 5A). I've not got around to reading the USB4 spec yet but doubt any of those are going away.

Thunderbolt 3 could work with the fastest of those but was limited to a maximum length of 0.5m unless using an active cable.

[patstew]

With USB 3.1 and 3.2 there were 6 types of cable with USB-C at each end. Three speeds (USB 2.0, 5Gb/s per lane, 10 Gb/s per lane) multiplied by two current ratings (3A, 5A). I've not got around to reading the USB4 spec yet but doubt any of those are going away.

Thunderbolt 3 could work with the fastest of those but was limited to a maximum length of 0.5m unless using an active cable.

That's true, but I maintain that for 99% of cases a consumer only needs to know the difference between the 2 I mentioned. It's pretty rare to need 5A or 10Gbps, and the cases that do are often a built-in cable to a dock or dongle anyway.
The extreme length restriction with thunderbolt 3 also applies to a lesser extent for 10Gbps, and cables that support USB3.x are fairly thick and inflexible. There's a fundamental tradeoff between fast/long/flexible/high current/cheap, and if you insist that all cables are fast and high current then they'll also be short, inflexible and expensive. Some people will prefer some other tradeoff, so we have different cables.
I think the way to improve the situation is OS notifications to inform you what the result of the negotiation was (E.g Charging rate limited to 3A due to USB cable, this device supports 5A charging")

[starshipeleven]

This is entirely incorrect.

USB-C cables as defined in the spec can be anything from "barely fit for USB 2.0" to "fully shielded and capable of carrying up to 100w", which makes the cable specification completely worthless in practice.

If the USB cables and equipment meet the USB specifications (including USB Type C, and optionally USB PD), then the cable will not catch fire or reach dangerous temperatures.

Yeah, good luck with that in the real life when most hardware is not really as compliant as you might think. Most chargers will happily overload (overheat end eventually shut down or burn out in a minute or so) if you attach a device that draws say 5A to a USB 2.0 cable. The only reason this does not happen that often (apart from the lack of devices needing so much power) is because the sink device has a smart charge controller that adapts to the cable's actual current capacity by sensing something. If the sink device isn't smart enough you can burn normal USB cables with ease.

Pushing all connections from simple and cheap USB 2.0 to complex and demanding like Thunderbolt in a cable with the same connector was a big fucking mistake as that will cause massive confusion and make it very easy to sell substandard garbage that is still technically legal for the spec, and we will get proof of that soon enough.

the cable is electronically marked

Yet another bad thing, as markers can be put anywhere. First few years they will be too expensive, probably but a few years in they will be dirt cheap. USB4 isn't going to last only a few years.

They should have required that the charged device should be able to test and react to the cable load/temperature to avoid issues automatically, like some are already.

Any cables carrying the USB trademark that do not actually meed the USB specifications are actually unlawful to sell in the United States and probably most other countries, as that is a trademark violation.

Good luck enforcing trademark violation on cheap USB cable manufacturers or importers.

And no I don't care that this allows them to dodge the liability. Manufacturers just making new cables with the USB logo on them regardless of spec are not a new thing, they should have kept that in mind too.

[starshipeleven]

Logic suggests someone is profiting from the situation, but I don't understand who or where. It isn't like a user can easily replace the USB on their device with a superior one (or one that does what they want), so how does the fragmentation generate profit?

Consumer buys more than one cable or dongle, which may or may not work. And already here we are looking at multiplying by 2 or 3 the profit.

Bad experiences with random cables will push the consumer to buy the more expensive ones, hoping that it will solve the problem.

[starshipeleven]

40 Gbps you say?

And how are we doing with ethernet these days?

10Gbit is VERY cheap if you go with SFP+ modules and fiber, and ridicolously expensive if you use RJ45 and ethernet cables.

Datacenters are using 40Gbit or even 100Gbit networking right now, so the technology isn't sitting around.

[edwaleni]

And create a bunch of new ones none asked for.

"is this cable good enough for going beyond USB 2.0?"
"is this cable actually able to deliver up to 100w?"
" is this cable actually a USB 2.0 cable with USB-C connectors so it's completely worthless for most intents and purposes?".
"What is this port even able to do?"

The moment where they unified the interface without providing OBVIOUS indication on the connector of what this cable or port is actually capable of, they started straying further and further from Zod's light.

(though to be fair, whoever invented the micro-USB for USB 2.0 and USB 3.0 belongs on a cross too)

Correct me if I am wrong Starship, but isnt USB4 just a protocol over Thunderbolt now? In fact all the backward support is just legacy device protocols over Thunderbolt?

The host chipset will support the transport and muxing and the pins will determine how it will be used. If the laptop has USB3 A ports hooked up the chipset will just mux and switch it.

If you plug in a USB-C cable and certain pins are ID'ed the new chip will train the cable and determine what services are available.

Intel already does this with their chipsets and USB support today. So I didnt see this as any great leap in technology.

The only gotcha I picked up was USB-C type connector cables today don't have all the needed pins to register USB4 support OR the cable has inadequate twist to support the higher speeds TB supports and will either throttle down or denote lack of features.

[Neraxa]

USB-C was a good idea to stave off 100 vendor specific cables for multifunction interfaces. Its better to add one connector type rather than 100 connectors for each vendor. Things have not become overly complicated, it has reduced complexity.

OVerall USB has been a huge success in reducing complexity and improving performance. Replacing the slower serial, parallel, PS/2, AT Mouse, joystick ports and so on with USB has dramatically decreased the amount of complexity with computer connections and improved performance. USB 4 is a vital upgrade to the standard to keep up with newer faster external storage.

[kylew77]

I wonder how long USB 4 support will take to trickle down to the 4 BSDs. NetBSD just got USB 3 support fairly recently. Would a device plugged into a USB 4 socket function at USB 3 speed without the driver or would it not function at all like the difference between USB 2 and 3?

[polarathene]

"is this cable good enough for going beyond USB 2.0?"

Some cables will have a SS USB 3.x logo near the connector or a printed label along the cable, others only in packaging. If none of that is present, assume it's the lowest tier. Not always the case, Samsung T5 SSD USB-C cable for example is just a short 0.5M black cable.

When buying a cable(rather than coming a cross one that you don't know it's capabilities), if it advertises USB 3.1/3.2, assume it's only supporting gen1, unless it specifically states gen2(10-20Gbps).

Cables that come with your products, use them with the products which they're known to work for, or test out the capabilities yourself and label the cable so that you can identify it for use elsewhere. If you have a variety of products and want a single cable that's inter operable between them, then buy one that's 3.2x2 100W capable and use that. If you want to use it with other USB devices, you can get small adapters or cables that have magnetic connectors for switching the connector types.

"is this cable actually able to deliver up to 100w?"

There are 2 power variants(3A/5A) and 3 data variants(480Mbps/5-10Gbps/10-20Gbps), for a total of 6 cables(perhaps 8 with USB4?).

You need a power source that can provide the amount of power you want to deliver, and can negotiate it over USB-PD(or alternative protocols like QC4), and of course a device that can request and receive that amount.

The cables are either 3A or 5A capable(so minimum capable of 15-60W depending on USB-C power source), power sources are often only able to offer up to 45W or less, some 60W, and fewer 100W. Some power sources are like USB hubs, in that they have a total power output, where if you use the other ports for power as well, you might not be able to get the supported max wattage over a port. You'll probably find more confusion/hassle with the power sources than the cables tbh when wanting 100W.

One aspect that'd be less obvious is voltage drops. If the cable is a high AWG size(the thinner it is, the higher the AWG size), it's not going to deliver the voltage from A to B ends well, this worsens over distance. So generally, a short thick cable(like Samsung T5 comes with) is ideal. That's especially the case if you want ThunderBolt 3.0 compatibility iirc(Passive kind). Otherwise you can go longer but need to keep in mind the thickness of the cable. I've seen some sell with 18 AWG size, for comparison, shitty USB 2/3 cables can be 28-32 AWG iirc, they'd not be great at bandwidth transfers, or delivering much power.

" is this cable actually a USB 2.0 cable with USB-C connectors so it's completely worthless for most intents and purposes?".

That's generally the cheapest kind of cable and is used for charging purposes, for which it can still do 3A. Some devices use this, and it might become a more common approach for charging, vs those 5V/2.4A at best USB cables before USB-C.

3A also needs to be negotiated for iirc, if the device or cable is not USB-C compliant(eg USB-A cable or adapter), and no Fast Charge Protocol(FCP from the various mobile vendors) is supported by the power source, then up to 1.5A can be delivered. Which is what Battery Charging(BC) 1.2 permits, or USB 3.2.

Some USB-C ports just provide USB 2.0 (Khadas VIM3 for example). So it's not pointless/worthless, your not likely going to buy one of these individually as someone who's tech savvy, but you might buy products that provide them... to be used with that product. If you'd rather pointless waste a more capable cable on such a device which cannot benefit from it, go nuts.

"What is this port even able to do?"

Valid question. Apple if I'm not mistaken is guilty of causing this issue already with USB-A ports being grey, so it's not clear what USB protocol version they support, and have shipped products with a mix of 2.0 and 3.0 ports iirc?

As a tech savvy consumer, you'd probably know what the port is capable of, provided the product specs indicate such. That said, just because we have clear USB-A 3.0 ports, doesn't mean we can be sure we'll get 5Gbps throughput. Even with UASP, an external USB 3 drive can only manage about 400MB/sec tops iirc, over gen2, it's different as not only does it bump up to 10Gbps, there's improved line encoding, reducing overhead.

But uhh, chipsets and various other joys that make USB often confusing/complicated, is already the case prior to USB-C. So technically this question isn't new, it obviously doesn't mean it's in anyway improved the issue, perhaps USB4 ports would be colour coded, or ensured to have some symbol(which is still required for certification by USB-IF iirc) which should indicate the support(but not necessarily indicate how capable of reaching peak performance the port is).

The moment where they unified the interface without providing OBVIOUS indication on the connector of what this cable or port is actually capable of, they started straying further and further from Zod's light.

The best way is to just buy the product, and if needed, label it in some way that you can differentiate/identify it from others if you have multiple. You can't really tell from a glance what hardware a machine is running either, capabilities aren't really limited to USB-C cables and ports. What sort of situation is being imagined here? Some foreign port you're unfamiliar with capability wise and/or you don't have the ability to use your own USB-C cable? That's really not unique to USB-C.

Here's some products:

100W gen2 1M cable
https://www.pbtech.co.nz/product/CAB...r-Delivery-USB

This one has support for Passive Thunderbolt too(The thunder bolt logo indicates it's ThunderBolt cable, and the number 3 indicates it's 3.0 afaik, specs state 32AWG though which sounds fishy, might just be for the USB 2.0 data lines only):

https://www.monoprice.com/product?p_id=24721

Anker 100W Gen2 1M:
https://www.amazon.com/Anker-Powerli.../dp/B072JYDQ7N

More available here:

https://usbcurrent.com/cables/

Note the Thunderbolt 3 note at the bottom about passive vs active(not desirable)