USB4 Version 2.0 to Offer 80Gb/s Transfer Speeds

[jaytv111]

Nope.
As the Windows PC world is transitioning to PCIe5, SSD controller manufacturers are on board. We are seeing M.2 SSDs with transfer rates exceeding 80Gbs introduced.

So - once USB4 v2 sockets are available, there will be external drives that push that transfer rate to the limit. The question is when the industry will make the transition. USB4 is still notably absent in the Windows hardware world, in spite of it being around for a couple of years. Unfortunately.

USB4 is present in any Thunderbolt 4 laptop and Thunderbolt 3 maintains intercompatibility with USB4.

So it's not so much as absent as it is unnecessary since most of the Intel machines use Thunderbolt.

The only machines that really needed USB4 were AMD machines that typically didn't implement Thunderbolt, though they could.

 

[R!TTER]

since most of the Intel machines use Thunderbolt.

They most certainly do not, a sizable number does support TB but it's not that big as compared to their overall installed base or for that matter the number of (new) devices they sell, the vast majority of them simply do not support TB.

 

[jaytv111]

They most certainly do not, a sizable number does support TB but it's not that big as compared to their overall installed base or for that matter the number of (new) devices they sell, the vast majority of them simply do not support TB.

I was speaking about the majority of Intel laptops that have been available in the type-C era (not overall installed base since that includes a lot of machines that predate USB-C). If you have a type-C equipped Intel laptop chances are pretty good it has Thunderbolt. There were many that did not have Thunderbolt, even Microsoft's own Surfaces for years just had plain USB-C. Dells, HPs, Lenovos, Alienwares, ones with Intel and USB-C have had Thunderbolt like 9 times out of 10.

 

[deconstruct60]

I guess they decided PAM-3 plus a 25% frequency increase would be easier to implement than "PAM-4".

But that would lose compatibility with existing passive cables and they are clearly motivated to keep that!

Passive cables are pragmatically correlated with shorter cables. If you take a really short, high quality Cat5 Ethernet cable and put it between two 10GbE ports then you'll probably get the speed throughput. Send through 50ft of Cat5 is likely trouble. The longer passive cables are probably on a slippery slope at the full "max passive" length defined by USB4/3.2 . The ones with cheaper build quality may not work so well, but work good enough at 40Gb/s. I think the USB-IF standards have a implied presumption that the cable makers are not cutting corners when define these speeds and max lengths. The USB-IF also tends not to hold implementers to high compliance standards either once in mass production.

If have just barely qualification for 20Gb/s cables is the noise floor going to low enough for PAM-4?

The cables are passive. Whatever the ports put on them is what goes through. I suspect will loose more "existing passive cables" on build quality issues rather than whether incrementally bump the frequency or require a lower noise floor on the cable.

So doubling the signal levels from 2 to 4 would be the only way of sticking to the same frequencies and thus to the same passive cables and still doubling the data rate.

It is not implausible that driver and adaptive receiver technology has advanced enough by now to make that feasible.

Advanced at what cost levels? One of USB-IF bugaboos is cost increases.

On a modern Intel system Thunderbolt needs a retimer just to get from the SoC to the external port. If the retimer costs go up many system vendors moan about the bill of material cost increase ...

 

[SactoGuy18]

Now you wonder is Apple looking at a new USB Type C port that no only supports USB4 Version 2, but also the new USB-IF Power Delivery 3.1 240 W charging standard simultaneously. Just this change could make it viable to consider getting rid of the MagSafe connector again.

 

[R!TTER]

I hate to break it to you but that's also not accurate.

I was speaking about the majority of Intel laptops that have been available in the type-C era (not overall installed base since that includes a lot of machines that predate USB-C). If you have a type-C equipped Intel laptop chances are pretty good it has Thunderbolt. There were many that did not have Thunderbolt, even Microsoft's own Surfaces for years just had plain USB-C. Dells, HPs, Lenovos, Alienwares, ones with Intel and USB-C have had Thunderbolt like 9 times out of 10.

I have an Acer laptop with USB C & no TB support, it's at best low double digits (percentage) even for the brands you mentioned.

 

[jaytv111]

I hate to break it to you but that's not also not accurate.View attachment 2050212
View attachment 2050212

I have an Acer laptop with USB C & no TB support, it's at best low double digits even for the brands you mentioned.

OK let me rephrase. The majority of the time, I FOUND that Intel USB-C equipped machines right now have Thunderbolt, which imparts USB4 support. There are plenty of low-end machines that skip Thunderbolt, and there were Microsoft Surface devices that skipped Thunderbolt for years, but mostly dealing with mid to high end machines they have Thunderbolt. You might find some brands that don't have Thunderbolt you might find some that do. It doesn't matter, I was saying just that USB4 is supported by the Thunderbolt machines out there which is a sizable chunk.

And I was just trying to say that that is why you won't see USB4 only on many of the Intel machines, you will see Thunderbolt 4 instead, and USB4 is not absent so much as branded differently MUCH OF THE TIME.

 

[ifrit05]

Just wait for USB4 Version 2.1 Mk III

USB4 Version 2.1 Mk III Gen 2x2

 

[R!TTER]

And I was just trying to say that that is why you won't see USB4 only on many of the Intel machines, you will see Thunderbolt 4 instead, and USB4 is not absent so much as branded differently MUCH OF THE TIME.

Yes & that's a really good thing considering you get the full fat USB4 experience with TB, not some gimped version having optional features omitted from the official USB4 spec. That's what I've observed with USB4/TB ports on Intel devices. At least Intel's doing that part right.

 

[deconstruct60]

They most certainly do not, a sizable number does support TB but it's not that big as compared to their overall installed base

If point at the Installed base most of them don't support USB 3.2 either. That isn't really Thunderbolt specific. It is far more so because they predate the standard. ( When Thunderbolt went USB-C port at version 3 is about where it more so started the real mainstream merge. It was mainly Apple pushing the rock up the hill before that. )

or for that matter the number of (new) devices they sell, the vast majority of them simply do not support TB.

New devices at any price point sure. But most of hose devices don't have USB 3.2 2x2 either.

Amazon's current top 20

Amazon Best Sellers: Best Traditional Laptop Computers

Discover the best Traditional Laptop Computers in Best Sellers. Find the top 100 most popular items in Amazon Computers & Accessories Best Sellers.

www.amazon.com

There are two chromebooks and a couple of other Celeron/Pentium systems there. [ back to school season so might be a skew there at the moment. ]

If filter out the sub $600 laptops and go a minimal Gen 11 Core processor (and up) then Thunderbolt is not uncommon. There is just one gaming rig in the top 20 that doesn't have it. Others do.

The folks way down the the super tight budget range of buying a $400 laptop top... are they even going to buy USB 4 device? They cost substantive money.

Thunderbolt being rare once get close to the nominal Mac pricing range isn't quite true. Where Intel has implemented the TB controller in the SoC ( starting at Gen 11 Core ) thunderbolt is not so rare. The system vendor has already bought the controller. The retimers don't cost that much. For Gen12 laptops not having Thunderbolt is actually even more rare.

upcoming X1 Fold Thunderbolt 4.

ThinkPad X1 Fold | Foldable PC, OLED Laptop

Get the ThinkPad X1 Fold from Lenovo. The worldâs first foldable PC. This revolutionary OLED laptop is perfect for work, travel, and entertainment. Shop now for free shipping & returns!

www.lenovo.com

New X1 Carbon

"... Lenovo gave the X1 Carbon a gentle overhaul for the Gen 9 model last year, swapping out the 16:9 screen for a 16:10 version and ditching Lenovo's semi-proprietary docking port in favor of a pair of plain-old Thunderbolt ports. ... "

Lenovo ThinkPad X1 Carbon Gen 10 review: A fast-but-flawed version of a great laptop

Intel’s 12th-gen CPUs make for a mixed upgrade to one of the best laptops.

arstechnica.com

Proprietary docking ports have been declining since Gen 11 arrived. The progression is only picking up speed with Gen12 systems.


Generally, Windows desktop systems are a boat anchor holding back TB adoptoin but those are not the majority of CPUs that Intel sell into the PC market . ( Gratuitously trying to loop in server sales has some problem of missing fancy modern USB features also. )

AMD taking more laptop share is another boat anchor.

But even with those two TB adoption on new systems has made huge progress over last 2 years.

 

[DEMinSoCAL]

Most certainly, we can be sure the cable will just have a USB symbol on it with no mention of speed, charge capability, data only or charge+data, display capability, etc. You know...it'll look like every other USB-C to USB-C cable you have.

 

[jsalda]

Well, technically they will end in tomorrow. To be exact - transfers will be instantaneous, but you will be able to see your data in the new place only the next day when reality catches up with transfer speeds. There is an international group of researchers under the WHO umbrella, who try to find ways how to avoid this from happening. Some say a USB cable might have to be plugged someplace into the body during the transfer.

What if the data is inverted? Will it then show up tomorrow, but you won't see it until yesterday?

 

[jsalda]

Who else remembers good 'ole lap link for MS DOS. 115.2k felt crazy fast back then. LOL

I remember when SCSI was the tops. Oh the days of switching ID numbers and cables until you get just the right combo and then you DO NOT TOUCH IT...EVER.

 

[KOTN91]

Why on earth would anyone transfer data using a cable. iCloud has been a thing since 2011

 

[dave070]

Wouldn't USB 4 followed by speed rating make more sense? USB 4 (80GB/s)?

 

[AppleFan1998]

You have to be a bit technically inclined to are what type of cable you bought. For most people they don’t care as long it works well enough.

 

[Damian83]

Lightning already charges about as fast as Apple probably wants (up to 27 watts). There are phones that charge at 50+ watts but it's not likely Apple wants to get there. USB-C iPhones I predict will charge up to 30 watts which is where the iPads with USB-C are at, only 3 more watts.

Battery health is okay with fast charging, they just make sure to check the temperatures and ensure the charging rate isn't high when the phone is hot. And they slow down charging as the battery percent gets higher.


Usually there are labels but you have to understand the labels. If they say SS and they have the USB logo, they are probably 5 Gbps USB 3.x gen 1 cables. If they say SS10, they are probably 3.x gen 2 cables (3.1 or 3.2, they are the same when gen 1 or gen 2). If they say SS20, then it's 3.2 gen 2x2.

Thunderbolt cables need the Thunderbolt logo: https://en.wikipedia.org/wiki/Thunderbolt_(interface)#/media/File:Thunderbolt_Symbol.svg

So if you see the Thunderbolt logo you will know it's a Thunderbolt cable.

The PD rating is split between 60 watts, 100 watts, and 240 watts cables. You pretty much have to look at the manufacturer's website, packaging, or some review somewhere to see its PD rating, or you just test it out yourself. A cable might also say on it how many watts it has. 240 watts cables are brand new and just came out and not too many devices use more than 100 watts for the time being so you can discount needing 240 watts yet.

The big problem is how you can't easily get all the features in one cable, including Thunderbolt, and you sacrifice something to get all the features or you pay a lot of money. Apple's passive .8 m Thunderbolt cable has all the features (and being passive supports the new 80 Gbps mode) but it's only .8 m, Apple's 1.8 m cable costs $130 and it appears it won't support 80 Gbps because it's active and not certified for the new 80 Gbps mode, same for the 3m cable which is more expensive at $160.

Apple it's not the only company in the world that makes cables (including TB) 🤦‍♂️
u can get a TB3 1,2m cable on aliexpress for 15$, that's less than a genuine lightning cable. and no, it doesnt explode, no, it doesnt break (at least not more than apple cable does), and no, it doesnt release toxic substances. But, yes, it doesnt fill Cook's wallet

 

[deconstruct60]

Wouldn't USB 4 followed by speed rating make more sense? USB 4 (80GB/s)?

That speed increase may not be required and isn't the only feature.

 

[theorist9]

Why does it have to be iso-frequency? What PAM-3 gives a path were they don't have to double the frequency.

You'll have to ask @constructor . I didn't think it needed to be iso-frequency, and indeed wrote this:

If the word length of a binary system is x bits, then that encodes for 2^x values. To determine the number of bits, y, that would give you the same no. of values in a ternary system, you'd just solve 3^y = 2^x, giving y = x * ln(2)/ln(3) = 0.63 x. I.e., you'd only need 63% as many bits, giving 1/0.63 = 1.6 times the bandwidth for the same frequency.
...
So, at least according to this back-of-the-envelope calculation, it sounds like they are going to have to increase the frequency by a factor of 2/1.6 = 1.25.

I guess they decided PAM-3 plus a 25% frequency increase would be easier to implement than "PAM-4".

Then it was constructor who replied that it did:

But that would lose compatibility with existing passive cables and they are clearly motivated to keep that!

So doubling the signal levels from 2 to 4 would be the only way of sticking to the same frequencies and thus to the same passive cables and still doubling the data rate.

It is not implausible that driver and adaptive receiver technology has advanced enough by now to make that feasible.

And since Thunderbolt of course must use the same drivers and receivers as USB it must be a common physical-level specification anyway.

Upping the frequencies would make it dicey to keep the same cables around – even with improved signal conditioning cables might become more critical than they would have to be.

So you should direct your question to them.

Having said that, USB's own language is unclear—when they say up to 80 Gbps using existing 40 Gbps passive cables, do the mean they're going to increse the frequency, such that some current 40 Gbps passive cables will handle 80 Gbps, and some won't?

https://www.usb.org/sites/default/files/2022-09/USB%20PG%20USB4%20Version%202.0%2080Gbps%20Announcement_FINAL.pdf

 

[constructor]

In principle it is possible that they're upping the frequency and try to compensate with better drivers and receivers so at least most of the same cables will still work, but sticking to the same frequencies and increasing the number of bits on each transition might make it easier to preserve cable compatibility.

The net error rate is the question in either case, though, and that is the critical point.

 

[theorist9]

I've always thought they should just name it by the speed: USB5, USB10, USB20, USB40, USB80, etc. Simple, clear, functional, and efficient. The idea is that that the name contains the specification. So you don't to have go to some table to find that, e.g., USB 3.2 Gen 2x2 is 20 Gbps.

And if they want to update, say, the 10 Gbps standard with added functionality, they can just call it USB10a, USB10b, etc. I.e., the 10 will always be there to tell you how fast it is. The letters afterwards tell you of any genertional update to the 10 Gbps standard.

But based on what I just read in this article by Scharon Harding of Ars Technica (https://arstechnica.com/gadgets/202...goes-where-no-usb-standard-has-gone-before/3/), it seems USB-IF has wised up to how byzantine their existing nomenclature is, since they're now suggesting this branding:

 

[cashboxz02]

Great. Now get new routers to use these instead of those ancient ethernet cables.

 

[Unregistered 4U]

I've always thought they should just name it by the speed: USB5, USB10, USB20, USB40, USB80, etc. Simple, clear, functional, and efficient. The idea is that that the name contains the specification. So you don't to have go to some table to find that, e.g., USB 3.0 Gen 2 x 2 is 20 Gbps.

Which would be great if USB ONLY transferred data. None of those designations would help if a person’s using USB for power delivery.

And based on what I just read in this article by Scharon Harding of Ars Technica (https://arstechnica.com/gadgets/202...goes-where-no-usb-standard-has-gone-before/3/), it seems USB-IF has wised up to how byzantine their existing nomenclature is, since they're now suggesting this branding:

Yes, but, most importantly, it’s optional AND requires certification. A company making a cheapy cable that they want on the market next week? NO label required!

 

[dave070]

That speed increase may not be required and isn't the only feature.

That makes sense. Thanks.

Their naming scheme isn't ideal though.

 

[profcutter]

I've always thought they should just name it by the speed: USB5, USB10, USB20, USB40, USB80, etc. Simple, clear, functional, and efficient. The idea is that that the name contains the specification. So you don't to have go to some table to find that, e.g., USB 3.2 Gen 2x2 is 20 Gbps.

And if they want to update, say, the 10 Gbps standard with added functionality, they can just call it USB10a, USB10b, etc. I.e., the 10 will always be there to tell you how fast it is. The letters afterwards tell you of any genertional update to the 10 Gbps standard.

But based on what I just read in this article by Scharon Harding of Ars Technica (https://arstechnica.com/gadgets/202...goes-where-no-usb-standard-has-gone-before/3/), it seems USB-IF has wised up to how byzantine their existing nomenclature is, since they're now suggesting this branding:

View attachment 2050381

View attachment 2050382

That would be like the old FireWire standard. Not a bad idea really. You knew if you were getting a FW800 cable, it could handle up to 800 Mbps if all the other components were up to the standard.