USB 4.0 [pdf]
usb.org
usb.org
This is a little more than USB 3.2 2x2 + TB3 but there's one crucial difference: the new, still optional USB 4 Gen 3 (what TB3 was) now carries USB packets as well. Previously TB3 only carried PCI Express and DisplayPort packets and the Intel hubs had USB root hubs in them. This solution worked out so well some PCIe enclosures opted to use two TB3 controllers just to get a problem free USB experience...
Also, PCIe was nerfed by Intel in TB3 for unknown reasons to 22Gbps, there's hope it'll be 32Gbps this time for real. You can still run a 3440 x 1440 @ 60 Hz monitor on the remaining bandwidth (or to spew marketing BS, 4k @ 30 Hz).
I'd just like to clarify, and add some info:
- you're not running the video over pcie
- ...unless you use an egpu (which has been a mostly problem free experience for me but wow YOUR MILEAGE MAY VARY)
- the "problem" with pcie-via-tb3 is not bandwidth, but latency. You can run "normal" GPUs on 4x with minimal penalty for most games.
- problem free USB experience... if you're using titan ridge with modern drivers - some laptop and enclosure manufacturers ship and include in their built-in updating software old drivers. Get the latest from intel and hope for the best. Alpine ridge is still buggy, and also quite limited in many ways. Fortunately, most laptops sold in 2019 shipped titan ridge. The only way to tell if you have alpine ridge is to count how many tb3 capable usbc ports you have, if the laptop you bought maxed out the usbc ports - which it might not have on an ultraportable. insane.
- you can only get the highest speeds (>20Gbps) over a 6inch cable or a USB Type-C Active Cable. Both sides of the cable include a chip which performs clock, sync, and some power negotiation of the link
My biggest issues with my eGPU have been with Windows' weird limitations on addressing space for devices, so if you have 3+ graphics cards (builtin primary, builtin secondary, eGPU card) you end up having to endlessly screw around with disabling and reenabling devices and device buses to change the order they're initialized in, messing with bootloaders to change which PCIe lanes get loaded at startup, etc etc. It's a nightmare.
By contrast, all my experiences on Mac have been either "it just works" or "after one reboot it just works". Still not ideal, but not anything close to the 4+ hours it took me to get it working on Windows for the first time (and after which it immediately stopped working again the first time I disconnected the Thunderbolt cable).
Now I use it with a Macbook Pro and it works very smoothly, even using a TB2 adapter. I can only use it in bootcamp because of Apple's policy towards nVidia, but it works every time and disconnects every time.
Even with the troubled experience, being able to move a GPU between computers is really just awesome, I wouldn't buy another machine without Thunderbolt or USB 4 eventually.
I have not had any of these problems. Intel GPU, Nvidia MX150, AMD RX 580. I wonder if there's some weird driver interaction stuff?
But yeah, the technology is clearly not at perfect maturity. All I can offer are condolences :(
Are they going to do that thing again where they rebrand the existing standards from USB 3 Gen X to USB 4 Gen X?
Kind of like the magnetic Apple charger for the older MacBook.
Why magnetic? So that I can quickly attach and detach a flash drive to it, without actually plugging it in. And then have an external mechanical locking pin or mechanism, to hold the device or cable in place, if you need to better secure it.
Why would this be superior?
* You eliminate the actual wear and tear of the USB metal housing.
* The external locking mechanism can be easily repaired or replaced when it wears out or breaks.
* It would probably help make your computer more waterproof.
* attracts everything magnetic and mettalic, not just the connector - need to be careful no to attract to many metal turnings, washers, etc. * the number of contact points & the quality of contact might not be enough for high speed communication * the contact pads/pins might be more exposed than normal connectors, so more likely to be damaged or accumulate dirt
USB 3.1 only has bandwidth for 4K @ 60hz w/o USB; so with non-thunderbolt cables (like these magnetic adaptors) it's either Power+4K@60Hz or Power+USB+4K@30Hz.
All 24 pins are connected, so video, data, and power all work as expected.
I would love one to. Except there wont be another port once EU pass the USB-C law.
I have used volta cables for a while (I preferred the 1.0 to the 2.0) and I'm surprised that this design idea isn't more widely popular, given how user friendly it is.
One only needs to glance at USB 3.0, 3.1 and 3.2, the rise of 4.0, ThunderBolt, the current cable disaster, etc, to appreciate why adequate standardization would be a net positive for the market (less friction, better adoption - benefiting consumers _and_ vendors), and provide for smoother industry adoption going forward.
I understand that USB-IF standardized all the plug formats. This was critical to get right in 1999, otherwise nothing would have interconnected properly and USB would never have caught on. If we are in a similar situation today, standardization of magnetic coupling is no less critical.
Beyond the technical factors, standardizing magnetic attachment critically prevents vendor lock-in by eliminating the value proposition for introducing unique and deliberately incompatible designs. Without standardization, there is currently unlimited potential behind establishing vertically integrations (walled gardens) around custom designs. A great example of where this can benefit vendors while fragmenting the market and creating terrible user experience is the charger industry, which the EU recently began to enforce stronger restrictions on to try to eliminate e-waste (https://news.ycombinator.com/item?id=22204174). USB-IF is likely familiar with this case.
Given the huge convenience factor and the practically infinite ways magnetic coupling attachments could be designed, it's plausible that USB might end up in a world of standardized plug-based connectors next to some n>1 quanity of vendor-specific approaches to magnetic attachment. (This is already happening, albeit as USB-IF ratified plugs next to arbitrarily-designed magnetic adapters built into cables.)
While I'm not especially familar with the technical details, I get the impression the market may carefully be circling Apple and their use of the MagSafe connector, and perhaps also related patents that may not yet have expired. I wonder if it's possible nobody wants to be the first to ask the question, for fear the answer capitalizes on the situation in a way that does not unilaterally benefit all. The only solution may be careful timing - waiting for successful market penetration, then quickly working to standardize before vendors get orientated.
(Clearance: feel free to quote and/or use for parts if/as you wish)
Standardise as much as possible.
• Two-lane operation using existing USB Type-C® cables and up to 40Gbps operation over 40Gbps certified cables
• Multiple data and display protocols that efficiently share the maximum aggregate bandwidth
• Backward compatibility with USB 3.2, USB 2.0 and Thunderbolt
In my household, USB Type-C has in practice been a great simplifier. Less so at work, but still not more complicated than the multifarious uses of, say, the modular connector family, or the DIN series.
I've gone from multiple chargers with different ports, to multiple chargers with the same port, and specific cables with tags on them, and having to explain to my partner that you can't plug x into y, even though they all have the same port.
* Devices are a 2015 MacBook pro, an external GPU, a Huawei phone, a pixel 3, a Nintendo switch, two sets of wireless earbud. For those, I have 4 specific chargers, and two specific cables (3 if you count the moulded plug to the Nintendo switch).
* Pixel - 9V/2A or 5V/3A by the looks of it. A decent USB C charger will do this
* Nintendo Switch - just don't use anything other than the Switch charger or a power bank for this. It's 15V/3A, but finicky.
* Macbook Pro - not familiar with this, didn't realise they went USB C all the way back in 2015, but it probably needs 60W+ and an e-marked cable.
* Huawei phone - if you give me the model number I can probably work this out. It's very hard to work out what that charging spec is, but it may also work with Qualcomm QuickCharge, which apparently makes the USB-IF very sad when delivered over the Type C cable.
I use a charger similar to https://www.kogan.com/au/buy/mbeat-gorillapower-80w-5-port-u..., and I also highly recommend something like https://satechi.net/products/satechi-type-c-power-meter-for-....
Now imagine explaining your comment to a lay person!
> it probably needs 60W+ and an e-marked cable.
I wasn't even aware e marked cables were a thing. I guess I need something similar for correct Android auto support with my phone and my car?
Honestly, your comment, while incredibly helpful, explains how unclear and how much of a mess it is.
P20 Pro: Uses "Huawei Supercharge" which is not Huawei Fastcharge or VOOC, nor is it compatible with USB PD or Qualcomm QuickCharge. Jesus. I'll cross that brand off my list completely
re: Android Auto - I have no idea, sorry. Double check the cable works correctly to transfer data to a PC (or for tethering etc) - the last time I looked, AA was VNC over IP over USB.
The dock can be fussy, but I think the switch itself is fine with anything that doesn't horrifically overvolt pins.
Using a generic power brick for the dock wouldn't have much impact. But changing the console itself would be a big loss toward charging it on the go.
It's very confusing to say to a consumer - hey, the cable that fits your phone won't actually charge your switch, even though if you switch cables, it will work with both. Except of course, your phone won't fast charge.
At least when my 3ds, Nokia and razr phones had different connectors, I wasn't going to be under the impression they would work with each other
I much prefer to just look at the cable and go 'nope, that clearly doesn't work with that' instead of playing cable roulette.
For all intents and purposes, they _are_ incompatible already, they're just pretending to be.
Especially the greater number of pins, the wider variety of possible uses.
But USB is only 4 pins.
IIRC when USB first appeared on consumer devices, Radio Shack was still going, but USB was the first consumer connector that they did NOT have a do-it-yourself solderable or crimpable cable terminal for, nor any truly appropriate cable. And they never did appear as years went by.
You had to cut your cables in half and splice them to appropriate extensions if you wanted to take advantage of the full 92 foot cable-length maximum under USB 1.
So glad that HDMI superceded that one.
What a weird design, especially for not even being 900.
> I have devices that can output video through their type c port but only when using a Display Link adapter which on paper needs USB 3 but the device only supports USB 2 data throughput.
They did nonstandard video output on all kinds of previous phone ports too.
But what part says you need to support USB 3 to support an alternate mode with video?
ok. truthfully, the sliding latch on the Ethernet AUI cut my fingers more. I hated that.
No, USB4 is a successor of Thunderbolt 3 (but similar enough that it can be backwards compatible with Thunderbolt 3). For instance, Thunderbolt 3 tunnels only PCIe and DisplayPort, while USB4 also tunnels USB 3.x.
USB4 is more of a Thunderbolt 4. Not only can it tunnel USB natively, it seems likely to be supporting a newer higher speed higher than TB3's 40gbps (but if it don't, I'm still happy with 40).