Thunderbolt 3 To Offer 40 Gbps Transfers, USB Type-C Connector

[runrun]

The 3rd problem is the friggin 100W supply. First of all, not all devices even have a power supply that can offer that much juice for even 1 port. That being said, I'm sure the vast majority of USB/TB ports won't even supply 25W.

Isn't it 100W for charging, only 15W for devices?

[dammarin]

Second, the average USB rechargeable device cannot handle that much power. The device itself should not be obligated to contain components to regulate the power and I personally would not trust the computer to control how much power it should be outputting. If the computer fails to properly recognize the device, it will either supply too much power and ruin the device, or, it will supply so little power that it won't charge. I don't like that kind of risk.

I think you're confusing watts with volts for a bit there.

[chithanh]

If I understand correctly, any combination of USB and Thunderbolt 3 cables/devices are mechanically possible, but only certain ones will actually work.

TB3 host <- TB3 cable -> TB3 peripheral: works
TB3 host <- TB3 cable -> USB peripheral: ???
TB3 host <- USB cable -> USB peripheral: works
TB3 host <- USB cable -> TB3 peripheral: doesn't work
USB host <- TB3 cable -> TB3 peripheral: doesn't work
USB host <- TB3 cable -> USB peripheral: ???
USB host <- USB cable -> TB3 peripheral: doesn't work
USB host <- USB cable -> USB peripheral: works (of course)

Because it is all USB Alternate Mode, charging will follow USB PD standard, i.e. up to 20V * 5A = 100W.

[schmidtbag]

I think you're confusing watts with volts for a bit there.

No, I am not. Quote from article:

Thunderbolt 3 also allows for 100 Watt charging and continues to support DisplayPort, PCI Express, and USB protocols over the same link.

lol and even if I were referring to volts, what kind of electronic DC devices are YOU using that operate at 100V?

[GreatEmerald]

If I understand correctly, any combination of USB and Thunderbolt 3 cables/devices are mechanically possible, but only certain ones will actually work.

TB3 host <- TB3 cable -> TB3 peripheral: works
TB3 host <- TB3 cable -> USB peripheral: ???
TB3 host <- USB cable -> USB peripheral: works
TB3 host <- USB cable -> TB3 peripheral: doesn't work
USB host <- TB3 cable -> TB3 peripheral: doesn't work
USB host <- TB3 cable -> USB peripheral: ???
USB host <- USB cable -> TB3 peripheral: doesn't work
USB host <- USB cable -> USB peripheral: works (of course)

Because it is all USB Alternate Mode, charging will follow USB PD standard, i.e. up to 20V * 5A = 100W.

Are you sure there's such a thing as a "TB3 cable"? My understanding is that it just uses the standard USB-C cable.

[chithanh]

GreatEmerald,

Yes, there will be three different kinds of cables. Passive copper, active copper and active optical.

Intel is offering a few different versions of the Thunderbolt 3 cable ? a 20Gbps copper cable that supports USB 3.1, DisplayPort 1.2 and 20Gbps Thunderbolt, and an active 40Gbps copper cable version as well. Intel is also working on a 40Gbps Optical Fibre cable as well for expected release in 2016.

http://thenextweb.com/gadgets/2015/0...in-30-seconds/

[peppercats]

External GPUs will finally be a valid option that doesn't require ridiculous specific laptops/hardware.
Can't wait to get rid of my desktop.

[dragorth]

It is a specification for a connector form factor and a description of what the pins for the connector can do. One of the things the pins can do is carry additional data, like video, but that fits within the normal specification of USB Type-C.

As far as I can tell, Thunderbolt 3 does not. It uses the same connector, but it doesn't sound like it fits within the existing USB Type-C standard.

The Type C connector specifically allows different "modes" for the USB 3 lines to operate in. This is the way it already handles DP and HDMI. If the device supports it. This just adds another mode, Thunderbolt 3.

[dragorth]

I have found statements saying a TB3 port on the computer side supports USB 3.1, but I can't find anything that suggests TB peripherals need to. Do you have a link saying that? Logically, it couldn't, since any device requiring faster transfer speeds or networking won't work with a standard Type-C port.

Thunderbolt 3 on the computer supports daisy chaining Thunderbolt devices. It can support daisy chaining these devices, and one USB 3/3.1 device at the end of this chain. This works the same way display port used to, the computer handles the logic, the peripherals just pass though the data. USB 2 has its own seperate lines that this does not affect, so you can plug them in as well.

The exception to this is if a device implements its own USB 3/3.1 and offers a pass through. Such a device is not limited in this way.

[dragorth]

Thunderbolt and USB 3 are getting dumber and dumber after every release. USB used to be a problem because some people (not including myself) had a hard time figuring out which was the right side up. The C connector is supposed to fix that, but now people are going to get confused about "is this Thunderbolt compatible?" and for people who don't have the C connector, they're going to get confused, thinking "well, this is a USB 3.x device, why doesn't it fit?"

The next problem is the speed of USB. 5Gbps was really pushing it, but it was doable since many ARM platforms and PCIe 2.0 compatible PCs could handle it. Anyway, the point of USB is universal compatibility, and they're totally ruining that when you buy a 10Gbps+ device and it pretty much only works (at full potential) on brand new intel-specific hardware. Considering the only [portable] devices that tend to need that much bandwidth are hard drives, I'd rather go with eSATA and save myself the insane amount of CPU load that comes with USB.

The 3rd problem is the friggin 100W supply. First of all, not all devices even have a power supply that can offer that much juice for even 1 port. That being said, I'm sure the vast majority of USB/TB ports won't even supply 25W. Considering how diverse the power supplies will be, no device will be designed to require 100W, so what's the point of going that high?
Second, the average USB rechargeable device cannot handle that much power. The device itself should not be obligated to contain components to regulate the power and I personally would not trust the computer to control how much power it should be outputting. If the computer fails to properly recognize the device, it will either supply too much power and ruin the device, or, it will supply so little power that it won't charge. I don't like that kind of risk.

Intel is trying to make a 1-size-fits-all and they're totally overcomplicating everything.

The first problem could be easy to fix. Just make all those ports TB 3. 😉

The second problem has been addressed in the newer USB specs. USB 3 and 3.1 offload a large amount of the processing to the controller, including the USB 2 and 1 processing. And USB 3.1 also adds new features for hard drive to support better access to make things faster.

Now, the line has to support bi-diretional 100W power. It doesn't have to provide it. Though in the case of desktops it most likely will. It may in laptops as well, for limited time frames.

And it is intelligent about the power it puts out. The spec calls for the device requesting more than the USB 2.0 power minimum. So, no problem there. And this is all the USB Type C spec, not TB3. They just adopted it.

Now, TB also provides the 15W mentioned before. I don't know if that is in aggregate.

Wow, you would think you believe these guys never designed a hardware spec before.