USB4 Specification(usb.org)
usb.org
USB4 Specification
https://www.usb.org/document-library/usb4tm-specification
314 comments
Reading through this, I'm left unable to see the underlying forces. None of the people working on this spec are idiots, so how did they arrive here? What are the forces and unstated requirements behind the scenes that I don't understand that have led to a USB4 spec with this many potential flavors and absurd naming conventions to express them? Is it something about cost? Or patents? Or some random company who won't budge on something but needs to be included?
The real crimes are:
1 - non-mandatory labeling. What does a given port support? What about a given cable?
2 - the name "USB4 Gen 3×2". Honestly I have never had the faintest idea what any of the various absurd names invented by the USB IF might mean. Superspeed? High speed? USB 3.1?
1 - non-mandatory labeling. What does a given port support? What about a given cable?
2 - the name "USB4 Gen 3×2". Honestly I have never had the faintest idea what any of the various absurd names invented by the USB IF might mean. Superspeed? High speed? USB 3.1?
When the USB-IF announced that they're going to incorporate Thunderbolt into USB4 I already worried that this might further complicate USB hardware and software.
Turns out: My worst nightmares just became true. We now have running USB over Thunderbolt. I don't even know where to start facepalming about this clusterfuck.
Take DisplayPort for example: Up to now we had the DisplayPort alternate mode for USB-C. That just worked for most devices with USB-C. Now we also have the tunneling of DisplayPort over Thunderbolt (as Thunderbolt-enabled Macs offer since Apple introduced Thunderbolt in its devices and which they use when connecting to Thunderbolt-enabled monitors). Both methods seem to be optional for USB now, so in future a host and a peripheral device might support "DisplayPort over USB", but while one might only supports the USB-C DisplayPort alternate mode, the other one might only support DisplayPort tunnels over USB4, causing them not to be able to talk DisplayPort to each other.
I guess I'm going to facepalm for a few hours now, while I read the specification in more detail.
Turns out: My worst nightmares just became true. We now have running USB over Thunderbolt. I don't even know where to start facepalming about this clusterfuck.
Take DisplayPort for example: Up to now we had the DisplayPort alternate mode for USB-C. That just worked for most devices with USB-C. Now we also have the tunneling of DisplayPort over Thunderbolt (as Thunderbolt-enabled Macs offer since Apple introduced Thunderbolt in its devices and which they use when connecting to Thunderbolt-enabled monitors). Both methods seem to be optional for USB now, so in future a host and a peripheral device might support "DisplayPort over USB", but while one might only supports the USB-C DisplayPort alternate mode, the other one might only support DisplayPort tunnels over USB4, causing them not to be able to talk DisplayPort to each other.
I guess I'm going to facepalm for a few hours now, while I read the specification in more detail.
As a consumer, can I please have a standard where I don't have to chase a new version every few years?
I know that many are backwards compatible, but they aren't forward compatible -- my MicroUSB cables don't help at all with USB-C, and type A and B connectors are still around.
By the time all of my micro USB devices have disappeared from my household, I'm sure a new thing will have emerged that's incompatible with USB-C, and so I'll still have to have several different cables around.
I know that many are backwards compatible, but they aren't forward compatible -- my MicroUSB cables don't help at all with USB-C, and type A and B connectors are still around.
By the time all of my micro USB devices have disappeared from my household, I'm sure a new thing will have emerged that's incompatible with USB-C, and so I'll still have to have several different cables around.
Please father Christmas may I just have one present?
A standard that is actually standardized in practice.
To this day I still don't know exactly how to buy a cable for my raspberry pi 4s. (Yes more a USB c than USB 3 issue but you get my point - a standard that isn't standardized fails at it's raison d'etre)
A standard that is actually standardized in practice.
To this day I still don't know exactly how to buy a cable for my raspberry pi 4s. (Yes more a USB c than USB 3 issue but you get my point - a standard that isn't standardized fails at it's raison d'etre)
Good. USB3 is still a mess, better make a new one before that gets sorted out.
The more I look at the USB mess the more I feel that a revised faster Ethernet with provisions for both insulated and direct connections (magnetics-less == cheaper chipset, lower power consumption etc) for board to board communications then a smaller/sturdier connector plus PoE built in would solve every problem USB tried to solve including the ones it introduced.
Apple Inc.
Hewlett-Packard Inc.
Intel Corporation
Microsoft Corporation
Renesas Corporation
STMicroelectronics
Texas Instruments
Edit: actually reading a bit further to the "Acknowledgement of Technical Contribution", this is the list of promoter companies. The contributors list is way longer and includes everyone you'd expect.
Why does USB have two seemingly unrelated versioning schemes? There's USB 1, 2, 3 and now 4 as well as Type A, B and C. This mess looks like manufacturers wanted to create a way to trick the average customer into buying the wrong cable or adapter, whereas initially the reason for creating USB was to increase compatibility.
Good job completely failing to address actual problems, USB-IF. What do people actually need? They need to be able to easily tell what the port on their device is capable of doing and they need to avoid buying chargers, cables, and devices that violate the specification, especially when they do so in a way that might destroy a device or even endanger the user. Unfortunately the only thing the USB-IF cares about is "Moar megabits! Moar!"
probably really dumb question with a really obvious answer that i don’t see, but...
why not have a packetized protocol like ip over the cable and devices just go over that? wouldnt things like enumeration, identification, connect/disconnect, interleaving protocols and encryption be much simpler since it’s a known domain?
wouldn’t that be more “universal” than cramming usb2, usb super speed, pci-e, displayport (which afaicr also wraps a few things) etc over a single cable?
am i missing something or is this just a cost issue...?
[edit] slightly answering my own question: pci-e is a packetized protocol... so i guess cramming everything in is more backwards compatibility to get all the other connectors subsumed into a single cable...?
why not have a packetized protocol like ip over the cable and devices just go over that? wouldnt things like enumeration, identification, connect/disconnect, interleaving protocols and encryption be much simpler since it’s a known domain?
wouldn’t that be more “universal” than cramming usb2, usb super speed, pci-e, displayport (which afaicr also wraps a few things) etc over a single cable?
am i missing something or is this just a cost issue...?
[edit] slightly answering my own question: pci-e is a packetized protocol... so i guess cramming everything in is more backwards compatibility to get all the other connectors subsumed into a single cable...?
Will this bring higher PD? A lot of laptops require 100W-130W of power (see, Lenovo X1E or Dell 7590), while the highest PD I've seen from usb is the Apple charger for 86W. I know Dell produces proprietary 130W usb-c charger, but I'm wondering if usb4 will help make it a standard.
I think USB is terrible. One problem is that USB addressing is not by the physical port that it is connected to (among other things, this is a security problem, but there are many other problems with this too); does USB4 correct that? Another problem is how vendor codes and that stuff is work; it should instead be identified by protocol type, and to allow the user to override the protocol type with the operating system functions.
There is other stuff is better, such as RS-232, and Compact Flash (unfortunately it lacks a write protect switch; SD has it but only the full size SD cards and not the smaller size; SD card has other problems though), and then also add some of my own specifications too, such as IMIDI and Digi-RGB, and you can have separate ports for each of these four thing.
There is other stuff is better, such as RS-232, and Compact Flash (unfortunately it lacks a write protect switch; SD has it but only the full size SD cards and not the smaller size; SD card has other problems though), and then also add some of my own specifications too, such as IMIDI and Digi-RGB, and you can have separate ports for each of these four thing.
The Serial Time Link Protocol (STLP) looks really cool to me - I've thought for awhile that the next-gen in audio gear would be more modular front ends that sit on top consumer protocols, which is where things are moving in broadcast/commercial audio installations today. I can imagine a whole host of consumer and professional devices that solve real problems at lower costs than we can today. Really interesting stuff.
edit: one weirdness, STLP has a recommended packet rate of 62.5kHz (actual quote "A Router shall periodically transmit a Serial Time Link Packet on TMU_CLK_OUT. The period between transmissions is implementation specific, but it is recommended that a Router transmit a Serial Time Link Packet every 16 μs"). If the only application domain (besides compliance testing) is audio, how did they arrive at that recommended rate, where it is rather different from standard sample rates (44.1k/88.2k and multiples of 8kHz)? I'm not sure how that affects clock synchronization.
edit: one weirdness, STLP has a recommended packet rate of 62.5kHz (actual quote "A Router shall periodically transmit a Serial Time Link Packet on TMU_CLK_OUT. The period between transmissions is implementation specific, but it is recommended that a Router transmit a Serial Time Link Packet every 16 μs"). If the only application domain (besides compliance testing) is audio, how did they arrive at that recommended rate, where it is rather different from standard sample rates (44.1k/88.2k and multiples of 8kHz)? I'm not sure how that affects clock synchronization.
Had to read the USB1 spec in 1998 to make a HID Device using a small micro, took me over 3 weeks to get it working and was debugging shit for months afterwards. Would never go there again and for future projects I decided ignorance was bliss and just used whatever stack the chip designers supplied, I suspect that covers 99% of all devices out there.
So basically, they are attempting to redo the mess that is 3.x
very cool.
"When configured over a USB Type-C® connector interface, USB4 functionally replaces USB 3.2 while retaining USB 2.0 bus operating in parallel. Enhanced SuperSpeed USB, as defined in USB 3.2, remains the fundamental architecture for USB data transfer on a USB4 Fabric"
very cool.
"When configured over a USB Type-C® connector interface, USB4 functionally replaces USB 3.2 while retaining USB 2.0 bus operating in parallel. Enhanced SuperSpeed USB, as defined in USB 3.2, remains the fundamental architecture for USB data transfer on a USB4 Fabric"
The Grrranimals standard worked. People would match up symbols to see what worked together. Why cannot technology do the same? Yes, the cables plug together, but one has a turtle icon and one has a car icon?
So USB 4 is basically Thunderbolt 3, got it.
(per Fig 2-1 "USB4/USB3.2 Dual Bus System Architecture")
Edit: with Gen3 speeds (20Gbs/link) and with optional Dual-Lane Bonding for increased bandwidth.
(per Fig 2-1 "USB4/USB3.2 Dual Bus System Architecture")
Edit: with Gen3 speeds (20Gbs/link) and with optional Dual-Lane Bonding for increased bandwidth.
USB3 and thunderbolt (unsure about their exact relation) enabled eGPU's, which is very cool.
What will USB4 enable?
What will USB4 enable?
Is this a universal serial bus, or a universal docking station port?
why the bloody fuck don‘t we get a new, clean thunderbolt bus — as it was invented by apple/intel — and leave the usb folks play their mindless games?
jesus.
jesus.
Why not standardise on USB-C and just make -C with the 2 added Thunderbolt pins (ala Apple) the norm? Why more specs
Why would they attach this as a zip file? Makes reading on an iPhone impossible (CORRECTION: cumbersome)
EDIT: Learnt something new. But the spec on an HTML page would be a lot more convenient than having to install more apps just to read this document.
EDIT: Learnt something new. But the spec on an HTML page would be a lot more convenient than having to install more apps just to read this document.