It doesn’t necessarily have to be just a magsafe 3 cable, as Apple gives you a usb-c to magsafe 3 cable that can handle 140 watts: 140W USB-C Power Adapter - Apple
Even though it’s not stated on the website, it’s assumed the current MacBooks (in atleast their magsafe connectors, the thunderbolt 4 connectors are also rated for USB4: so maybe?) support PD 3.1 up to 140 watts.
But I think the real issue is how new the technology is, probably not enough time for manufacturers to get PD 3.1 docks out.
Works with various models (today I’m using a Lenovo), BUT, relies on a power pin to get the full output.
Another option, which has worked for the HP Zbooks, is multiple power sources through different USB ports. I plug a 120w Zbook into two power-delivery USB-C docking monitors and it draws from both, getting enough to charge this way.
Not sure if the FW will be able to copy that trick. Also, you have to have two power sources. Funny solution but make of it what you will.
I think I recall Framework staff saying that simultaneous power input from multiple ports is not supported. If multiple are available, it will select the one with the highest power.
Higher wattage docks will be available in time. Until then, you can use a second cable if you need more than your dock provides (I know, two cables, how will we live).
Yeah, but you know it’s not that. It’s about getting the power adapter out and connecting that up every time. That’s a nuisance. Alternatively of course just buy two power adapters… one for the bag, one for the desk. Cheaper than a dock at least, not ideal, but works “for now”.
This conclusion is missing information. USB4 is modeled on TB4 however it actually more closely follows TB3. Every TB4 host supports USB4. Same cannot be said for USB4.
Minimum voltages are different for one. PCIe requirements are not same. TB4 supports 40Gb/s up to 2M cable. USB4 only up to 1M cable.
You are correct. I misread a review I saw for the TS4 that discussed 2.5Gb/s limitations when using USB4. This won’t be an issue for FW Ryzen implementation.
Thanks for pushing me to do more research on this.
That’s not quote right, TB4 literally runs on USB4, though it has stricter requirements and none of that “optional feature” bs the usb-if tends to pull. Though so far all the USB4 implementations I know of are full featured.
TB4 does have a few extra tricks like more dedicated DP channels and stuff that is mostly necessary for apple stuff since macos refuses to support MST for some reason.
That is straight up just the stricter testing thing, has nothing really to do with the technology used.
As far as the cable length thing, there’s already active cables that go beyond 2M so I suspect there will be manufacturer tested passive cables longer than 1M. However it will be use at your own risk when exceeding the spec, unless 20Gb/s is acceptable to the user.
I think you need to do a bit more reading. TB4, unlike TB3 is no longer a protocol, but only a certification for USB4 & other USB functionality. Every part of TB4 that you get on the connector is already part of some kind of USB-spec. Only not all parts are mandatory. How to tunnel PCIe through USB4 is defined in the USB4 spec. There are no minimum speeds, because all of PCIe tunneling is optional. TB4 just certifies that that optional feature is actually supported with at least 32G / x4 Gen 3. That is all.
And you are wrong about the cable lengths. Active USB4 cables can do 40G over 2m. USB4 is where the requirement came from that USB4/TB4 cables need to be backwards-compatible to USB3, DP and everything else. Also why USB4 allows greater distances with the same quality of cable. A TB4 cable is just a branded, more expensive and arguably better tested USB4 cable.
That is also how framework could just call there devices USB4 before they attained TB4 certification. It is all USB4, it has always been. Even the backwards compatibility to TB3 is part of the USB4 spec. If you actually check the Windows USB4 drivers or Linux drivers on any modern TB4 device like our Frameworks, they will distinguish only TB3 and USB4 as both, the hardware generation as well as the connection mode. There is no “TB4” happening on the cable. Going forward, that is purely the Thunderbolt Logo, marketing and certification efforts from Intel on top of stuff that already follows the USB4 spec.
Edit: I looked into your Ars article from further down and sadly, they are simply wrong on a few details. Looks like not the best source when it comes to USB-stuff and the disitinction between TB4 and USB4.
Edit2:
Here an excerpt of Intel’s public 13th gen CPU specs page 115:
TB4 is just a “solution brand” that requires the combination of the listed features of USB (including USB4) and DP features. It is not a protocol like TB3 was.
That is a very interesting term. I think that is more designed to obfuscate for example AMDs lack of the 2nd DP tunnel and we really should push manufacturers to just list the actual speeds for each protocol and tunnel. Most consumers might not care about that, but to call that full featured when TB4 supports more is misrepresenting. And even TB4 is not at the outer limits of USB4. ASMedia has shown you can put more than the x4 Gen 3 connection through USB4 (40G) and Intel and AMD even support that even though it is even beyond TB4 requirements (only newer integrated controllers of course).
HDMI and DP and USB themselves already have those positions (i.e. DO NOT advertise anything with HDMI 2.1, but instead list the Gbit/s supported, the modes TMDS/FRL and all the optional features). It is just nobody is following those rules and as a result consumers get more confused then they would need to be by that already quite complicated topic.
And then they try to instinctively assign some kind of meaning to stuff like “DP 1.4” or “USB 3.2” which will come back to bite them eventually / has already with USB3.
And we can already see the beginnings of people trying to shoehorn in USB 4.1 or USB 4.2 because they did not understand why and how they were already wrong when doing this with USB3.
Honestly better to just get rid of the differentiation entirely and just mandate everything. No optionals but unfortunately the usb-if doesn’t have the balls or the authority to actually do that.
And that devolved into an absolute mess too. Technologically having lots of options is great and what the usb-if has done there is amazing but unfortunately it turns into a total mess in prtactice.
Mhh. I of course would like that and am willing to pay for that. But I think the average customer might not agree with you and like stuff like USB-C hubs based on Vias VL830 that is just USB4 without TB3 compatibility and hence cheaper than any TB4 solution…
We are getting to a point it includes that many features, that mandating every single one, always, would often drive up the cost, waste silicon, because it is not usable in practice or disincentivize its presence on anything but the most premium products where price does not matter at all…
Mhh. It is a mess, yes.
But I think you can absolutely see what the USB-IF was trying to do.
The Spec versions were never intended for consumer use. They always had consumer-facing labels and logos for their speeds: FullSpeed, HighSpeed, SuperSpeed, SuperSpeed+ SuperSpeed+ 20G. Those existed from the beginning, were on every package, yet barely used in reviews, press, device specs etc. Those that used the numbers instead, including basically all publications did not realize what kind of dead-end this was heading to.
You can argue that they were to slow to react with that rebranding to only the actual speeds and that they screwed up, because for a year they did not clarify whether “USB 20G” meant solely USB4 20G or also USB3 20G (they are now clarifying only USB4, so USB3 20G even got its name axed). And yet, it is still not used in practice…
Calling the USB4 spec actually “USB4 version 1.0” was clearly designed to make it as hard as possible for idiots to continue to list the spec version instead of the proper name because it was shorter, but instead default to calling it “USB4” and the speed like intended from launch. Because they realized, even though never intended and communicated by them that way, that 99% of customers, reviewers and manufacturers just said “USB 3.1” to mean sth. it does not, instead of the always clearly communicated “USB3 SuperSpeed+”.
If anything, I would blame them for apparently not having the PR division to reach out to larger publications to get them to publish explainers and use proper names going forward. Using the spec versions is just not tenable for complex standards that live longer than 2 generations.
And Intel shows that they can just relable sth. as a premium-tier at higher price to solve exactly that problem for the consumer that does not want to concern themselves with the details and is willing to pay more in order to have all the compatibility expressed in a single term.
Will a thunderbolt dock work with the discrete GPU on the 16? My desktop will pass the GPU though thunderbolt but only if I connect a cable from the GPU to the MB DP input. I presume this would work but the Framework 16 is sort of an odd duck so I’m a bit worried a dock would only work with the iGPU.
The dock will connect to the iGPU, however the operating system will still be able to use the dGPU for most of the work and use the iGPU just for the final outputting to the display. This typically results in ~10% less performance than if the dock was connecting to the dGPU directly.
This is the case with the vast majority of laptops.
What GPU the dock connects to is controlled by the USB4/TB4 controller.
The USB4/TB4 controller built into any current CPU* only supports using the iGPU, so if Framework (or any laptop manufacturer) wants the ports to connect to the dGPU they would need to use separate controllers from the CPU. This is possible (IIRC Dell has done it on some 17-inch laptops), however separate controllers tend to have higher power consumption and worse performance than the controllers built into the CPU.
Furthermore Framework is already near the limits of the PCIe bandwidth capabilities of the current CPU in the Framework 16. This isn’t a problem for the integrated controllers (since those pull from their own dedicated separate bandwidth), however would be a problem with separate controllers. So it would likely also be necessary to switch to a different CPU, likely one with worse power efficiency and a worse iGPU.
* - IIRC some of Intel’s 11th Gen and earlier CPUs had integrated Thunderbolt controllers that could use the dGPU.
So almost no laptop will connect a dock to the dGPU. Framework or not.
From what I’ve read, there is a USB-C alt-mode port on the 7700S expansion bay itself which would connect to the dgpu directly. It would not be able to passthrough USB or charging via the dock with that port. I could be wrong though.
I believe that would only output DP video, since it’s basically a DP port just in a USB-C connector (DP alt mode). I don’t believe it will have anything else like USB, PCIE tunneling, or TB. Basically it’ll act like a USB-C connector out of a desktop GPU.
