With USB4 and USB 3.2, USB is more powerful and convoluted than ever

In recent years, the USB Implementers Forum (USB-IF) has developed this bizarre habit of releasing specification updates that are often tremendous steps forward but are obscured by confusing, misleading, or downright terrible naming. With the announcement of the USB4 (no space!) specification, the group has reached the apex of powerful features/confusing nomenclature.

USB4 is a big deal 

First, let us bask in the big news of the day: The USB-IF announced USB4, with no space between the "B" and the "4," cutting against all other USB naming schemes. Partially, that's no doubt because it's actually a new architecture, built on Intel's Thunderbolt protocol. Even so, it's compatible with Thunderbolt 3, USB 3.2, and USB 2.0. When an inferior-spec device is connected, a USB4 port will negotiate the "best mutual capability" of the two connected devices.

Thunderbolt 3 has been one of the (optional) crown jewels of USB Type-C, providing a blistering 40 Gbps theoretical bandwidth compared to USB 3.1 Gen 2's 10 Gbps. Now, the speed of Thunderbolt 3 becomes part of USB because Intel has allowed it to happen. Previously, companies implementing TB3 had to pay royalties to Intel because it was a proprietary technology. Now it's in the hands of the USB Promoters Group, thereby "enabling other chip makers to build Thunderbolt compatible silicon, royalty-free," according to an Intel announcement. Devices will still need a special controller on either end of the connection to take advantage of those delicious speeds, though. Not for nothing, upcoming 10-nm Intel chips will have native TB3 support. 

Also, USB 3.2

USB 3.2 has been kicking around since 2017, but recent stirrings at MWC 2019 indicate that we should be seeing the technology actually land on shipping devices soon. The presence of USB 3.2 has also further muddied the naming waters of the USB ecosystem. Like past USB 3.x updates, this one subsumes all of the changes and errata of the previous iterations, but it's big feature is a doubling of bandwidth.

USB 3.1 Gen2 can hit 10 Gbps, but it does so with a bus that has one Tx and one Rx; USB 3.2 uses a dual-bus architecture to double that to 20 Gbps. From the USB 3.2 specification document: "One bus is a USB 2.0 bus (see Universal Serial Bus Specification, Revision 2.0) and the other is an Enhanced SuperSpeed bus (see Section 3.1). USB 3.2 specifically adds dual-lane support."

Because any new USB 3.x subsumes the earlier USB 3.x specifications, the USB-IF prefers to categorize not by number (eg, USB 3.2, USB 3.1, and so on) but by speed. Looked at one way, that makes sense; USB 3.2 Gen 1 is 5 Gbps, USB 3.2 Gen 2 is 10 Gbps, and USB 3.2 Gen 2×2 is 20 Gbps. The "2×2" indicator makes it clear (ish) that we're talking about the two-way architecture, but you could make the argument that just calling it "Gen 3" would be simpler. Alas. 

Ports' importance

Let us not forget that titles like USB 3.2 and USB4 don't necessarily tell you what physical ports (eg, USB Type-A or USB Type-C connectors) are in play, nor other capabilities like Power Delivery. So, consumers need to understand that although technically "USB 3.2" runs over USB Type-A or USB Type-C, you'll only get the full performance of USB 3.2 Gen 2×2 over a USB Type-C cable. 

Essentially, then, the emergence of USB 3.2 and USB4 mark a huge push away from USB Type-A ports towards USB Type-C ports. The future is a two-way USB street; USB 3.2 gives us 20 Gbps and USB4 gives us 40 Gbps over USB Type-C. USB Type-A is stuck at 10 Gbps, no matter what name you slap on it. 

That has ramifications for the way companies built their products. Presently, manufacturers have often been loading up their devices with both USB Type-A and USB Type-C ports to accommodate both legacy devices and more powerful ones. It's likely that will continue for years to come, but perhaps with an increasing percentage of USB Type-C to USB Type-A. But it seems the days of USB Type-A ports are numbered.

Comments closed
    • Rakhmaninov3
    • 6 months ago

    UNIVERSAL lololololz

    • Grigory
    • 6 months ago

    Can we have a clear cut at some point in the future? Please?

    We could have USBfiVe (at least the name should be right down their alley) with optical data lines and USB-L (for legacy) for everything before it. Something like that.

    • psuedonymous
    • 6 months ago

    We get the usual “numbers are confusing!” comments here, but do you really want your motherboard (or phone, or anything else that has USB host capability) to need discrete USB1, USB2 and USB 3 controllers? Or do you want whatever the latest USB standard is to subsume the preceding standards? Because that’s [i<]exactly what happens[/i<], and why updated USB standards incorporate the lower speed ratings of previous standards while also adding the non-speed features of the newer standards (e.g. Type C support). Equating USB spec support with data transfer speed is a lazy shorthand, but not actually correct.

      • Laykun
      • 6 months ago

      I don’t think that’s where the confusion is coming from here …

    • NovusBogus
    • 6 months ago

    This is what happens when consortiums exceed their shelf life.

    • nerdrage
    • 6 months ago

    USB-IF, go home, you’re drunk.

    • tipoo
    • 6 months ago

    Thinking TB3 being integrated into USB 4 can only mean TB4 around the corner. TB3 still does limit GPUs and especially so if it has to loop back around to an internal laptop display, 4 doubling that bandwidth should be coming pretty close to native performance. There was a rumor it would debut with the stacking mac pro concept.

    • K-L-Waster
    • 6 months ago

    If you got a million monkeys and let them fiddle with a million interconnect protocols, eventually you would get the one, perfect, universal standard.

    In the meantime, we’ve got this.

    • NTMBK
    • 6 months ago

    So with Gen 2×2, the Universal Serial Bus is now two parallel buses?

      • K-L-Waster
      • 6 months ago

      One’s positive and the other is negative, apparently.

      • Captain Ned
      • 6 months ago

      Only if it’s Ms. Frizzle and the Magic School Bus.

      • Pwnstar
      • 6 months ago

      “Serial” parallelism?

      • Mr Bill
      • 6 months ago

      Bi-binary?

    • NTMBK
    • 6 months ago

    So with Gen 2×2, the Universal Serial Bus is now two parallel buses?

      • Pwnstar
      • 6 months ago

      No longer serial!

    • albundy
    • 6 months ago

    i just thought they were special needs and could not count past 3. considering the evidence, i still think its true.

    • ltcommander.data
    • 6 months ago

    Thunderbolt 3 required active USB-C cables for lengths longer than (~0.5m) which were not compatible with USB 3.1. Being based on Thunderbolt 4, does USB4 still require active cables for longer distances or otherwise address this backwards compatibility issue? Otherwise, if the USB 3.2 naming scheme wasn’t complicated enough, customers are going to have to deal with figuring out why their longer, active USB4 cables don’t work with their USB 3.2 and lower devices yet their shorter, passive USB4 cables do. And if they need long cables they’re going to have to maintain separate active USB-C USB4 and passive USB-C USB 3.2 cables that otherwise look the same.

      • adisor19
      • 6 months ago

      Finally, someone asking relevant questions.

      Adi

      • slate0
      • 6 months ago

      I’ve run into similar issues with cord length. I own 4 USB type c cables and I’ll have to do some real sherlock work if I run into a compatibility some day. I’m actively AVOIDING these new standards and the cost investment of buying new cables. I could buy a 6 ft cable of medium capability, but it’s not worth it, I’ll just move my hardware closer instead.

      The naming conventions are a ___show of advertising over functionality and simplicity, and someone needs to step in and put a stop to this junk. USB 3.1 and onward work DESPITE the work of the USB-IF.

    • uwsalt
    • 6 months ago

    Given the naming scheme dumpster fire that is being actively fueled by the USB-IF, I’m starting to think it would be easier for consumers and better for Thunderbolt’s adoption if the standard were named, marketed, and controlled by some other entity. Does Intel have a standard “No, you morons, give it back” clawback provision in its agreement with USB-IF?

    If that means Intel has to continue charging a royalty to cover costs of maintaining the spec and licenses — what, like $0.25 or even $1.00 per port? — then so be it. For even a marginally more coherent naming scheme –and even Intel’s impenetrable processor numbering would be an improvement — that’s money well spent, IMO.

    • Wirko
    • 6 months ago

    After USB 3.1, I was expecting 3.14, then 3.141, and so on, until we arrive at a well-known irrational number. I’m very satisfied with the current solution; both 3.2 and 4 are irrational enough.

      • ronch
      • 6 months ago

      After 3.1 it should be 3.11, not 3.14. That’s just shrewd of you to skip 3 version numbers.

        • Redocbew
        • 6 months ago

        I’d call it transcendental.

        • derFunkenstein
        • 6 months ago

        USB 3.11 for Workgroups

          • K-L-Waster
          • 6 months ago

          Ohh!, will it have a 32 bit network redirector?

    • chuckula
    • 6 months ago

    Put in some Thundercats references!

    • Waco
    • 6 months ago

    The only consistency in USB naming is that it’s inconsistent as hell.

    • Srsly_Bro
    • 6 months ago

    Just in time for this.

    [url<]https://www.techpowerup.com/253224/new-thunderclap-vulnerability-threatens-to-infect-your-pc-over-thunderbolt-peripherals[/url<]

      • Captain Ned
      • 6 months ago

      I thought you could only get the thunderclap in Bangkok …

        • Srsly_Bro
        • 6 months ago

        Same concept. If you plug in your peripheral unprotected, you’re getting the thunderclap.

        • Pwnstar
        • 6 months ago

        Is that where you go to get your kok banged?

    • davidbowser
    • 6 months ago

    What I want, and think would be incredibly valuable to just about everyone, is a standard for USB connectors and cables that does away with the concept of mixed capability. What I mean is that it should not be allowed to have a cable with connectors for USB-C and USB-A, but then only support data and power at USB 1.1 ratings. Bottom line is that in order for manufacturers to participate and have the logo, they should adhere to a higher minimum baseline of capabilities since anything less pisses off consumers and undermines the market and whole standards process.

      • Vhalidictes
      • 6 months ago

      My understanding is that this is already true – but most manufacturers (and users) don’t care if the cables are actually certified.

      Most certified cables aren’t even labeled as such.

    • drfish
    • 6 months ago

    I can’t even link to the obligatory XKCD because these are barely “standards” anymore.

      • K-L-Waster
      • 6 months ago

      There are now 15.2 standards with varying levels of quasi-compatibility.

    • Vhalidictes
    • 6 months ago

    I guess Intel made the decision to make TB more popular in lieu of trying for another profit center.

    Of course, that hasn’t and won’t work, as IBM has taught everyone repeatedly (rest in peace, MCA adapters and Token Ring networking)

    Now that TB is a standard part of USB, hopefully it will be cheaper and therefore succeed at being used. There’s no guarantees, since USB 3.0 is still around and still LCD, but at least now there’s a chance…

      • Questar
      • 6 months ago

      Intel announced in 2017 they were going to open TB.

    • qmacpoint
    • 6 months ago

    Dual USB 3.2 would’ve made MORE sense

      • Pwnstar
      • 6 months ago

      That’s hardly “serial”, though.

        • Prion
        • 6 months ago

        UPB 1.0

          • Grigory
          • 6 months ago

          Me and my printer are ready.

    • Neutronbeam
    • 6 months ago

    USB WTF?! Do we really need 3.2?

    • Growler
    • 6 months ago

    USB4 is now the ultimate power in the universe. I suggest we use it.

      • Redocbew
      • 6 months ago

      Now witness the power of this somewhat fragmented and bizarrely named revised protocol?

        • Neutronbeam
        • 6 months ago

        Is it fully operational?

          • curtisb
          • 6 months ago

          May I fire when ready?

            • maxxcool
            • 6 months ago

            sure juts buy a cheap cord off ebay and charge your multi-amp phone 😉

            • K-L-Waster
            • 6 months ago

            You’ll probably need an adapter, but when you find one, sure.

        • ALiLPinkMonster
        • 6 months ago

        Now I’m gonna picture tiny Imperial engineers pressing buttons and turning knobs and watching a green laser go through a tunnel inside my computer whenever I plug in a USB device.

          • willmore
          • 6 months ago

          What have you been doing until now?

          *shakes head*

        • just brew it!
        • 6 months ago

        Market at will, vendors!

      • Krogoth
      • 6 months ago

      The power of an ultra-high bandwidth I/O is insignificant to the power of the shill.

        • chuckula
        • 6 months ago

        If you thumb me down now I’ll become more trollerful than you can possibly imagine.

          • Mr Bill
          • 6 months ago

          Give in shill, to the power of the 4th USB protocol.

          • Krogoth
          • 6 months ago

          “Chuck, I’m your trollmaster…..”

            • K-L-Waster
            • 6 months ago

            “I’ll never bi-directionally handshake you!”

            • Krogoth
            • 6 months ago

            “Only if you knew the power of the backwards compatibility!”

          • Srsly_Bro
          • 6 months ago

          I got thumbed down for making a news post. Less likely my followers are here and instead some people really got offended at negative thunderbolt news.

          • derFunkenstein
          • 6 months ago

          There’s -3 buddy, prove it.

            • JustAnEngineer
            • 6 months ago

            The first rule of downthumb… is that you don’t talk about downthumb.

            • derFunkenstein
            • 6 months ago

            He asked for them!

            • Redocbew
            • 6 months ago

            The second rule of downthumb is: you don’t talk about downthumb.

            • derFunkenstein
            • 6 months ago

            The third rule of down thumb is: someone yells “troll!”, goes straw man, spews obscenities, the debate is over.

            • Redocbew
            • 6 months ago

            The fourth rule of downthumb is: There’s only [s<]two[/s<] one side to the debate.

            • chuckula
            • 6 months ago

            Cinebench 20 that’s actually properly coded with Embree raytracing for rendering was officially released within hours of your post.

            [b<]YOU HAVE UNDERESTIMATED THE POWER OF THE TROLL SIDE FOR THE LAST TIME[/b<]

            • Krogoth
            • 6 months ago

            [b<]"YOUR FEEBLE POWERS ARE NO MATCH AGAINST THE POWER OF SHILLING!" "IF YOU DO NOT TURN, THEN YOU WILL BECOME GLUED"[/b<]

            • Mr Bill
            • 6 months ago

            Your midichlorian count will be insufficient for ray-tracing UNLESS YOU TURN TO THE DARK SIDE.

      • Taxythingy
      • 6 months ago

      This is not the USB 3.2 Gen B enhanced++ cable you are looking for…

      • ronch
      • 6 months ago

      The question is, can we handle this much power??

      • Bobs_Your_Uncle
      • 6 months ago

      [i<]I am altering the protocol naming convention. [b<]Pray I don't alter it further![/b<][/i<] ... no, seriously ... PRAY. At this point, divine intervention may be our only hope ... Obi Wan Kenobi not withstanding.

        • drfish
        • 6 months ago

        This is the best post in this thread.

        • JustAnEngineer
        • 6 months ago

        This deal… is very fair, and I’m happy to be a part of it.
        [url<]https://www.youtube.com/watch?v=WpE_xMRiCLE[/url<]

      • Mr Bill
      • 6 months ago

      On USBchlorians we can rely, from now on.

      • freebird
      • 6 months ago

      isn’t it the last component needed to make the “Infinite Improbability Drive” work???

      Yea!!!!

        • Krogoth
        • 6 months ago

        Only version 4.2

      • cynan
      • 6 months ago

      Hear hear. Alas, the age-old problem of OEMs racing to the bottom conflicting with coughing up the extra scratch for more expensive control chips, cables, ports (probably) and extra bandwidth required.

Pin It on Pinterest

Share This