Photo by denvit on Pixabay
USB 3.2’s Naming Convention Is a Hot Mess; USB4 Announced
The upcoming USB 3.2 standard will provide a maximum speed of 20 Gbps when products supporting it hit the market later this year, but due to confusing labeling from USB-IF, the USB industry group, it may be difficult to figure out what speed any particular device supports. That’s because USB-IF is retroactively subsuming the older USB 3.0 and USB 3.1 standards under the new USB 3.2 name. Here’s how the new names work out:
|Spec||Speed||New Spec Name||Marketing Name|
|USB 3.0||5 Gbps||USB 3.2 Gen 1||SuperSpeed USB|
|USB 3.1||10 Gbps||USB 3.2 Gen 2||SuperSpeed USB 10Gbps|
|USB 3.2||20 Gbps||USB 3.2 Gen 2×2||SuperSpeed USB 20Gbps|
To be fair, USB-IF is being consistently confusing, since the initial release of USB 3.1 was called USB 3.1 Gen 2, and USB-IF also then retroactively renamed USB 3.0 to USB 3.1 Gen 1. What was wrong with the simple USB 3.0, 3.1, and 3.2 names?
The practical upshot of all these naming changes is that any given “USB 3.2” device might not provide the speed you expect, so you’ll have to read the specifications carefully.
Just before we published this article, the USB Promoter Group announced the draft specification of USB4, due in the middle of 2019. USB4 is slated to be backward-compatible with USB 2, USB 3.2, and Thunderbolt 3. In fact, it’s based on Thunderbolt 3 and doubles USB 3.2’s 20 Gbps top throughput to 40 Gbps, enabling it to support multiple simultaneous data and display protocols. What’s not clear is how USB4 will differ from Thunderbolt 3, given its backward compatibility, USB-C connector, and need for new cables certified for 40 Gbps.
I wonder if one of the differences between USB4 and TB3 will be the way TB3 is more directly connected to the system bus. USB has historically been more isolated. If you want the cutting edge speeds of TB3 it makes sense to feed it more directly, but as progress moves forward maybe the USB4 spec can use the bandwidth of TB3’s transport without needing to be so tied-in to justify it (and that could also reduce how much you need controller chips which have driven up the cost of implementing TB3 devices).
Just an interesting (to me) thought.
In a Pollyanna world, we’ll get fast storage via USB4 without paying the TB3 premium.
Since Intel opened up TB3 to be the basis of USB4, maybe they’ve got a TB4 waiting in the wings with 100Gbps throughput.
One can always hope.
Maybe that 100 Gbps TB4 standard could rely on a really sturdy connector like Lightning.
(I’m only partially joking)
I would like to learn more about the details. Is USB4 going to be essentially TB3 without the royalties (i.e. a new USB with DMA and giving up the host-slave model) or is it still going to rely on a host with a host controller and impose limits on the slaves like today’s USB does?
I guess for us in Mac land it won’t matter a whole lot since Apple will likely stick to TB, but if USB4 essentially is TB3 without the royalties I assume we could be looking at much more selection in terms of peripherals and probably lower cost gear too. $300 TB3 docks vs. $80 USB-C docks come to mind.
And here I thought keeping track of SCSI IDs was a pain…
Join the discussion in the TidBITS Discourse forum