USB-IF: USB4 Products Coming in Late 2020

69 Comments

View All Comments

• Lord of the Bored - Thursday, June 13, 2019 - link

  The security argument on IEEE 1394 basically boiled down to "DMA is bad".
  Ultimately, some tradeoff has to be made between usability and accessibility. We COULD just disable all access to the system, but... it wouldn't be very usable at that point.

• valinor89 - Wednesday, June 12, 2019 - link

  I am eagerly awaiting the upcoming USB 4.0 Gen 2x2x2 devices! AKA MegaSpeed USB 40Gbps to make it ore understandable.
  
  /s

• s-plus - Wednesday, June 12, 2019 - link

  By late 2020 Intel will release Thunderbolt 4 and USB will have the inferiority complex again...

• Arsenica - Wednesday, June 12, 2019 - link

  If USB-IF had any sense they would rebrand all USB type-C based versions as >4.0 versions, for example:
  USB 3.0: USB type A based connectors at up to 10 Gbps
  USB 4.0: USB type C based connectors at up to 20 Gbps
  USB 5.0: USB type C based connectors at up to 40 Gbps

• TheUnhandledException - Wednesday, June 12, 2019 - link

  "If USB-IF had any sense" ok you can stop right there.

• repoman27 - Wednesday, June 12, 2019 - link

  Quite sensibly, when the USB-IF publishes new versions of their specifications, they increment the version number of the document. Also, quite sensibly, they have always advised vendors not to use these version numbers to indicate the signaling capabilities of their products.
  
  The media thought it would generate a lot more clicks if they suggested that the situation was intractably complicated and that the specification version numbers were actually some sort of brand or trademark that the USB-IF had changed somewhere along the way, which is not at all the case.
  
  The USB has consistently registered silly marketing names for vendors and the media to use to communicate the various signaling capabilities of products to customers (Basic-Speed, Hi-Speed, SuperSpeed, SuperSpeed USB 10 Gbps, SuperSpeed USB 20 Gbps). Everyone universally refuses to acknowledge these as acceptable. So the USB-IF said that if vendors insist on using the specification version number on product packaging, advertising, or other marketing materials, that they would also have to indicate the signaling capabilities of the product. That's where you get the Gen 1 or Gen 2 (5 or 10 Gbit/s) for USB 3.1 products, and the x 1 or x 2 (single or dual-lane) for USB 3.2 products.
  
  We've always had USB Type-A and Type-B connectors, so I'm not sure why Type-C is so challenging for everyone. Once again, the media is partly to blame for repeatedly ignoring the USB-IF's messaging that Type-C is a cable and connector specification and is entirely separate from any particular version of the USB specification, which is what describes the signaling protocol.

• Lord of the Bored - Wednesday, June 12, 2019 - link

  Because version numbering is a convenient way to track capabilities. My older systems don't have USB 3.2 Gen 1 ports, they have USB 2.0 ports. Anything compliant with version 2.0 of the spec is a USB 2.0 device. That future versions of the spec are backwards-compatible doesn't mean older versions stop existing, or that there's no valid reason to reference them.
  
  And imagine the confusion if people DID start following the USB-IF suggestions and solely using cute names for things. Sooooo many shady devices advertising they were high-speed or even full-speed to sucker in the masses. How do you explain to your parents that full-speed is the slowest option, and that high-speed is pretty slow?
  
  The USB-IF is laying out a stupid and confusing naming scheme that only benefits dodgy manufacturers. Better that devices are labelled by the newest version of the spec they comply fully with.

• repoman27 - Wednesday, June 12, 2019 - link

  Older versions of the specifications do cease to existing as the USB-IF deprecates them. You can't download older versions of the specs from their site anymore, and they stop certifying new devices against them. It's just like software versions. However, devices that were designed and/or certified for previous versions do not cease to exist. The currently active versions of the USB specification may be 2.0 and 3.2, but billions of USB 1.0, 1.1, 3.0 and 3.1 devices continue to exist. And of course you can continue to use those version numbers and assume that people will know what you're talking about.
  
  However, say I design a new device targeting the USB 3.2 specification (a version that I can currently license) but it only requires 5 Gbit/s signaling (which is totally allowed according to the spec) and follow through with getting it certified according to the USB 3.2 testing procedures. It is dead to rights a USB 3.2 product. The USB-IF can't tell me as a licensee who has jumped through all their hoops that I can't refer to it as a USB 3.2 device. They can however tell me that if I refer to it as USB 3.2 on packaging or in advertising or other marketing materials that I also have to tell you which USB 3.2 signaling mode it's actually capable of, namely 5 Gbit/s. This is where the vendors are playing fast and loose and customers can potentially be mislead.
  
  The "cute names" which everyone ridicules are registered trademarks controlled by USB-IF, and as such they carry a legal mechanism to enforce usage. And the speed classes, as they currently stand, are Basic-Speed, Hi-Speed, SuperSpeed, SuperSpeed 10 Gbps, and SuperSpeed 20 Gbps. The only thing my parents might not understand about these names is what heck a Gbps is, and that SuperSpeed without a qualifier only has 5 of them. And I can hear my mom asking, "But how do I know how many Gbps I even need?" No solution is going to work for everyone.
  
  A significant percentage of what are called USB 2.0 devices that are on the market right now are only capable of the low-speed or full-speed signaling modes introduced in USB 1.0, rather than high-speed which was introduced in USB 2.0. And almost nobody cares because devices generally adopt whatever signaling mode is most appropriate, and backwards compatibility is guaranteed. This is pretty much the same deal for USB 3.x devices. If you actually need 10 Gbit/s vs 5 Gbit/s signaling or dual-lane operation to reach 20 Gbit/s, you're not my parents and you can probably take the time to read the fine print or do some research. The place these capabilities probably matter the most is in regards to host ports. I'm guessing 90% of people only care about whether a device works or not, not about the theoretical bandwidth the physical channel can provide to the upper layers.
  
  And you have it exactly reversed. The dodgy manufacturers know that if their device fully complied with the USB 3.0 spec, it is now also fully compliant with the USB 3.1 and 3.2 specs. People who look at spec sheets are more likely to buy whichever product has the bigger number listed. This is exactly why the USB-IF said you can't do that unless you also indicate the maximum signaling capabilities, and explicitly provided at least three different acceptable ways of doing so.

• Lord of the Bored - Thursday, June 13, 2019 - link

  I don't have it backwards. The current approach of the USB-IF makes "dodgy manufacturers know that if their device fully complied with the USB 3.0 spec, it is now also fully compliant with the USB 3.1 and 3.2 specs" true.
  3.1 and 3.2 should only be applicable to devices using one of those new features. That they are not is nothing more than the USB-IF doing manufacturers a favor so they can conceal their devices' interface functionality.
  Extensions of the specification should exist as a "sublevel" designation, like the IEEE does(most obviously with 802.11). And hey, for a while the USB-IF was doing good with that. Then they decided to "clarify" everything by declaring that USB 3.2 was ALL USB EVER, and created a "gen #x#" nomenclature to replace the existing "USB2, USB3, USB 3.1" designations. (And 3.1 should have been USB4, but that is minor nitpicking relative to the Gen X clusterbomb).

• edzieba - Thursday, June 13, 2019 - link

  There's a lot more to the USB spec than just link bandwidth. A naive "bigger number means more faster" does nothing to address all the other things USB need to do to be useful, which is why the new standards exist in the first place. Power delivery is one of the more obvious ones.