[mananaysiempre]

They are all USB-C as far as the physical interface goes, but their protocol capabilities are different: the rear ones are USB4 (actually full TB4 AFAIU, except Intel refuses to certify AMD systems), the front ones are USB3.2+DP and USB3.2. The fundamental reason seems to be that a “mobile CPU” is more of a SoC, including built-in USB capability, and (unlike the Intels) the AMD ones only do two USB4 ports, which on the Framework are routed to the rear.

This, then, is done through USB4-capable retimers, which turned out to draw noticeable power when a USB-A card with a pulldown is connected and idling; but no better retimers could be substituted. The front USB3.2 ports use different retimers and don’t suffer from this. (Why do the Intel motherboards not have this problem? No idea, but if I had to guess, Intel probably makes the retimers for those and is refusing to let them be used with AMD processors.)

[jychang]

That’s pretty cool information.

Also this is the type of stuff where Apple would make sure the end user would not have to worry about it. Charitably, they’d put in the money/effort to make sure all 4 USB ports are the same (notably, all USB ports on a macbook pro can charge at 100W, which is not a trivial task); or uncharitably, they’d just say fuck it and ship a laptop with only 2 USB ports.

[GloriousKoji]

> Also this is the type of stuff where Apple would make sure the end user would not have to worry about it

Some of the earlier macbooks with USB-C had overheating problems if the USB-C power cable was connected to any of the ports on the left side.

[viraptor]

Not just overheating, but overheating the chips to the level where they put extra performance stress on the kernel for some weird reasons. Basically I was seeing 70% CPU usage plugged in on the left and 20% usage on the right with the same task. MacOS has many issues like that, the "Apple cares more about users/design" idea is a bad meme at this point.

[AnthonyMouse]

It has been like this for years. Apple does some things well, and there are reasons to like those things, but their following is overzealous to the point of kneeling before what is, fundamentally, still a for-profit corporation.

I feel like business schools have failed this generation of management. They look at Apple and see them do something that customers like, but it has a cost to the company, like giving up the ability to track people. So they conclude that it will be an advantage if they defect and make money from ads.

Then they see Apple do something customers hate and get away with it and the lesson they take is that they can get away with it too. But they can't, because the reason why Apple got away with it was by doing the thing the customers liked. And then they can't understand why they can't achieve Apple's margins.

You idiots, do the opposite. Then you can charge Apple's margins from the first one and take their market share with the second one.

[KolenCh]

IIRC this is not a reflection on actual CPU use. Those extra CPU usage is more like throttling. I don't remember where I read this though, worth searching about this if you're interested.

[mananaysiempre]

I mean, yes, Apple (even before they get to making their own silicon) can slam a wad of cash on the table and say “make us a better retimer or else” or perhaps even “make us a better SoC or else”. Not a lot of other laptop manufacturers can. (Now that I’m looking, it seems that Intel and Kandou are basically it when it comes to USB4 retimers, and the Kandou ones—whose deficiencies we’re discussing here—only recently arrived on the market.)

That said, I believe all Frameworks will accept 100W down any of the four ports, even if the batteries are only designed to charge at 55 or 61W[1] and the rest will have to be consumed by the rest of the system somehow.

[1] https://community.frame.work/t/how-fast-can-the-battery-be-c...

[Aaron2222]

> and the rest will have to be consumed by the rest of the system somehow

The wattage figures on a charger are the maximum it can provide. The connected device only draws what it needs.

[mananaysiempre]

Of course. I just wanted to be clear that Framework laptops have 100W-rated power circuitry that works through any of the four ports, yet it would not be entirely correct to say that they “charge at 100W”: the batteries charge at 1C, that is capacity/1h; any power over that only counts towards powering the running system.