M2, Thunderbolt 5?

[deconstruct60]

Thunderbolt was developed by Apple and Intel.

Thunderbolt (interface) - Wikipedia

en.wikipedia.org

Thunderbolt is a trademark that Intel held by themselves.

Apple can't make Thunderbolt "work" by themselves. They don't sell components so that other folks can get things down in the Thunderbolt ecosystem. Do they get some dribbling of royalties? No. The underlying core tech of Thunderbolt has been contributed to USB-IF.

Pragmatically at this point what Intel is doing now is using the Thunderbolt brand to remove some of the "optional" stuff that USB 4 leaves up to the implementors. Tighter security; not optional . Video out on each port ; not optional . etc. etc. etc.

Same thing is likely to happen on next iteration of Thunderbolt . Alt Mode DisplayPort v2.0. ( that could be non optional in TBv4.5).


It is unclear how Intel could unilaterally go to PAM3 encoding and not have wide spread 'buy in' from USB-IF. Getting huge buy-in from USB-IF isn't likely going to happen soon at all. ( look at long gulf between USB 2.0 and 3.0 )

 

[deconstruct60]

Found out today Thunderbolt 5 is a thing.
Any word on this being part of the M2 series?
I ordered an ultra but if they release the M2 version one year later with double the external data-rate (80Gbps), it's worth waiting for.

Thunderbolt 5 is likely pretty far out into the future. DisplayPort v2 on a Thunderbolt "4.5" socket is likely coming before a version '5' arrives.

Going from 40Gb/s bidirectional to 80Gb/s unidirectional isn't a huge leap. Take the "inbound" lanes and turn the direction the data is traveling while trading off some flexibility.

 

[throAU]

Funny, hope that pushing the tech doesn't add more cost to buy. Barely any devices at Thunderbolt 3 on market. Many are still stuck in 5Gbps/10Gbps USB 3.2 Gen 2×2 and USB4 Gen 2×2/USB4 Gen 3×2 is just around the corner.

I am hoping USB4 to move out that mess of USB3.x naming. God dam confusing trying to figure out what type of cable I have.

It will of course cost more, not sure why anyone would think it won't - pushing 80 gigabit over copper cable isn't trivial or cheap.

But... there's not a heap of devices pushing for thunderbolt 3 yet as most devices simply do not require the bandwidth and can save money using cheaper older interface.

High speed NVME SSD, external GPU and maybe a few other niche cases like high speed camera storage is about it.

 

[EntropyQ3]

Thunderbolt is a trademark that Intel held by themselves.

Apple can't make Thunderbolt "work" by themselves. They don't sell components so that other folks can get things down in the Thunderbolt ecosystem. Do they get some dribbling of royalties? No. The underlying core tech of Thunderbolt has been contributed to USB-IF.

Pragmatically at this point what Intel is doing now is using the Thunderbolt brand to remove some of the "optional" stuff that USB 4 leaves up to the implementors. Tighter security; not optional . Video out on each port ; not optional . etc. etc. etc.

Same thing is likely to happen on next iteration of Thunderbolt . Alt Mode DisplayPort v2.0. ( that could be non optional in TBv4.5).


It is unclear how Intel could unilaterally go to PAM3 encoding and not have wide spread 'buy in' from USB-IF. Getting huge buy-in from USB-IF isn't likely going to happen soon at all. ( look at long gulf between USB 2.0 and 3.0 )

Or for that matter USB 4, control chips were out two years ago. Uptake in the industry is still almost zero. ?
(Hopefully that will change with the new generation of PC platforms coming within a year. Hopefully.)

It’s clear that Intel intends TB5 to maintain the USB-C connector, and thus of course maintain backwards compatibility with USB 4 and TB4. I’d like to see an upgrade happen. But then I’ve wanted 10Gb Ethernet to happen in consumer space for well over a decade, and I’m still waiting.

Sometimes the inertia of this industry, combined with market segmentation, is quite frustrating.

 

[deconstruct60]

Or for that matter USB 4, control chips were out two years ago. Uptake in the industry is still almost zero. ?
(Hopefully that will change with the new generation of PC platforms coming within a year. Hopefully.)

Except the chips pragmatically didn't really come. There was early talk that ASMedia was going to put out competitive chips. Looking over at their website all I see is USB 3.2 gen2 chips.

Fall of 21 Via talked as though has something "big" in USB4 .. but not really.

Via Labs Launches USB4 Controller, Paving the Way for USB4 Hubs

Cheaper USB4 hubs incoming

www.tomshardware.com

"... The company plans to start shipments of its VL830 to select partners in Q4 2021. If all goes well, expect availability of USB4 docking station and multi-function adapters in early to mid 2022...."

And while they mention "Hub" there in the announcement it really isn't a USB4 hub.

"...VIA Lab’s VL830 is a USB4 Endpoint device controller, featuring an optimized cost-structure and full compliance with USB4 specification. VL830 offers both USB and DisplayPort functionality and operates at full performance when used with Thunderbolt™ 4 or USB4 systems, and is also backward compatible with previous system that support DisplayPort Alternate Mode over USB Type-C, and features integrated USB 3.2 SuperSpeed USB 10Gbps Hub. ..."

https://www.via-labs.com/product_show.php?id=114

Technically it is not a USB4 hub (which requires supporting Thunderbolt) but a USB4 Endpoint(which happens to have a USB3 Hub. a v3 Hub doesn't make it a v4 Hub. ).

So pragmatically all the USB4 "hubs" are still completely reliant on Intel's Goshen Ridge (JHL8440) device controller .

Teardown of the Caldigt TB4 hub.

CalDigit TS4 Thunderbolt 4 Dock Teardown

Caldigit’s latest dock has excellent power specs and a huge number of ports, but lacks a couple features.

dancharblog.wordpress.com

All the Gen 11 mobile Intel processors supported USB4/TB4. So does Gen12 (Alder Lake ) and 13 (Raptor Lake) mobile. Technically, Apple's M1 doesn't qualify for TB4. The M1 Pro and Max got better number of video out support to make the spec. But the bulk of M-series SoCs shipped so far are the M1's. (and Intel is still skipping USB4 by default on desktop SoCs. )

Supposedly, AMD is going to support USB4 on next iteration, but technically they too could skip Thunderbolt 3 ability and still get a USB4 icon/branding. ( Again the only thing which closes loophole of optionally skipping TBv3 is a "USB4 Hub". Hosts and Enpoints can skip it to save a buck. ).


The expectation that there was going to be a rapid , substantial price decrease in USB4 hub pricing was probably misplaced. There are solutions to choose from. There is no "race to the bottom" market pricing forces though (either now or in next year or two).

It’s clear that Intel intends TB5 to maintain the USB-C connector, and thus of course maintain backwards compatibility with USB 4 and TB4. I’d like to see an upgrade happen. But then I’ve wanted 10Gb Ethernet to happen in consumer space for well over a decade, and I’m still waiting.

It makes very little practical sense for Intel to push on to TBv5 when most of the market hasn't fully digested TBv4 yet. Or even USB4 to a pretty large extent ( still not on Intel desktops and AMD is still lagging. ). Even Apple to some extent, the M1 powered iPad Air specs out at USB 3.1 gen 2. (probably to save some power and to enhance the market segmentation to the iPad Pro. ).

Thunderbolt 1 2011
Thunderbolt 2 2013 (+2)
Thunderbolt 3 2015-16 (+3 to fully rolled out) [ Titan ( v3+ was in 2018).
Thunderbolt 4 2020 (+4 )

The more folks involved , the longer the interim period is likely to be. That is a major contributing reason why USB-IF moves at a relatively snails pace most of the time. Tightly coupling "Thunderbolt" to USB-IF roll out momentum is likely going to slow the pace even more.

As pointed out above , no one else but Intel has come out with a real USB4/TB4 hub controller implementation yet. Apple has done their own TB implementation, but they aren't selling to anyone ... so who else? If Intel goes off an relatively quickly "obsoletes" TB4 then where is the incentive for a broader set of implementers? ( similar 'dust up' issues happens on USB3 when Intel was ready to go faster than everyone else and lots of folks wanted a more level playing field so that had choices and broader ecosystem).






Can get 10Gb Ethernet on more than half of the desktop Mac line up at this point. ( Mini , Studio , Mac Pro) . Intel's modern baseline chipset supports 2.5GbE which is at least some progress.
Part of he 10GbE adoption was getting simple unmanaged switch pricing down to what prosumers/consumers expected. I think there was lots of inertia there not to give up on 10GbE switch market up for a long time. And lots of consumer devices are leaning into WiFi (which is close to good enough ... when the whole household isn't on concurrent high-def videoconferencing sessions. ) .

Sometimes the inertia of this industry, combined with market segmentation, is quite frustrating.

It isn't just inertia. There is a dual edged sword of USB ecosystem is that the customers are price anchored to the long lived, mature tech also. If it costs too much then the USB-IF standards committee pragmatically has to put the feature into the "optional" category to get the new tech technically adopted. But since it is optional the implementers shooting for "cost effective" will drop the features and still get a USB (n+1) label ... just without the extra feature "adjectives".

 

[EvilMonk]

It would make more sense to add Wifi 6 or HDMI 2.1.

Unless Ape gives us TB 5 eGPUs

I have wifi 6 in my mini m1 and MacBook Pro 16” m1 Max

 

[EvilMonk]

Thunderbolt 5 is Intel's technology, they will only license it to third party manufacturer like AMD & Apple after releasing it first on their hardware, which they haven't yet.

Nope, it was designed by Intel AND Apple as a replacement for FireWire…

 

[edanuff]

Don't use Intel controllers in the Studio Display? Apple's TB controller is embedded in the M-series. The teardowns of the Studio Display have generally not gone down to the chip level, but I don't recall anymore announcing that Apple had a discrete TB controllers of their own.

Very disappointed with iFixit's teardowns recently, this used to specifically be the kind of thing they would cover but now it's all about the YouTube video and no longer the high-res pictures of the components. That said, at around 3:17 of their YouTube video there is what appears to be a JHL7440 Thunderbolt Controller on the Studio Display logic board.

 

[EvilMonk]

Very disappointed with iFixit's teardowns recently, this used to specifically be the kind of thing they would cover but now it's all about the YouTube video and no longer the high-res pictures of the components. That said, at around 3:17 of their YouTube video there is what appears to be a JHL7440 Thunderbolt Controller on the Studio Display logic board.

I imagine they still need the JHL7440 (Titan Ridge) I added a specs sheet of all Intel Thunderbolt controllers. The JHL7440 is probably needed to provide compatibility with Intel macs and PCs? since Apple implementation of TB isn’t exactly up to the Thunderbolt standards and the Studio Display can work on PC

 

[edanuff]

I imagine they still need the JHL7440 (Titan Ridge) I added a specs sheet of all Intel Thunderbolt controllers. The JHL7440 is probably needed to provide compatibility with Intel macs and PCs? since Apple implementation of TB isn’t exactly up to the Thunderbolt standards and the Studio Display can work on PC

That's a safe assumption. It's interesting that they didn't support HBR3 with the monitor since that chip can support it.

 

[MrGunny94]

Funny, hope that pushing the tech doesn't add more cost to buy. Barely any devices at Thunderbolt 3 on market. Many are still stuck in 5Gbps/10Gbps USB 3.2 Gen 2×2 and USB4 Gen 2×2/USB4 Gen 3×2 is just around the corner.

I am hoping USB4 to move out that mess of USB3.x naming. God dam confusing trying to figure out what type of cable I have.

Yeah I'm just putting some stickers on mine to know what's TB3/4

 

[theluggage]

Beginning 3 years, Intel opened up Thunderbolt licensing to anyone, i.e. Apple. So Apple doesn’t have to use Intel’s chips.

That's true up to Thunderbolt 4. It's up to Intel whether they do the same for any future versions of Thunderbolt.

Or for that matter USB 4, control chips were out two years ago. Uptake in the industry is still almost zero.

...because demand in the industry is low. USB 3.1 (whether on type C or type A connectors) is good enough for most low/middle-range uses. 3.1 is fast enough for all but the fastest external disc drives, for example, and only very fast external SSDs even bother with 3.2gen2. There's been little uptake of 5k or larger displays - and slow uptake of DisplayPort 1.4 which could drive them without Thunderbolt/USB4 (again, whether it is via USB-C connectors or DisplayPort). Anyway, I fail to see the point of combining data & video on a single connector when you are dealing with modern high-res and/or high framerate displays that eat most of the bandwidth.

USB 3.1 is ubiquitous because it is included in virtually everything from the cheapest PCs and Chromebooks upwards - and the huge economies of scale keep it cheap and the low-end markets are price sensitive. Thunderbolt/USB4 is only needed by a minority - and they also have the choice of modular PCIe systems with internal high-speed expansion - if you're using a PCIe GPU it has DisplayPort/HDMI outputs maybe USB-C alt mode - USB4 is an unnecessary complications.

If PC users do go for USB4/Thunderbolt because they need super-fast external expansion then there's an incentive to go with Intel Thunderbolt controllers for compatibility (ISTR that "legacy" Thunderbolt compatibility is optional in the USB4 spec),

Apple have gone out on a limb by building everything around Thunderbolt with a limited number of alternative ports - and pricing their only PCIe system up into the stratosphere. Maybe it makes sense for Apple's target market (I'm not going to revive that argument) but the rest of the PC market is unlikely to follow their lead when the bulk of customers aren't even at the limits of USB3.1 yet and affordable modular PCs are still a thing.

 

[EntropyQ3]

That's true up to Thunderbolt 4. It's up to Intel whether they do the same for any future versions of Thunderbolt.


...because demand in the industry is low. USB 3.1 (whether on type C or type A connectors) is good enough for most low/middle-range uses. 3.1 is fast enough for all but the fastest external disc drives, for example, and only very fast external SSDs even bother with 3.2gen2. There's been little uptake of 5k or larger displays - and slow uptake of DisplayPort 1.4 which could drive them without Thunderbolt/USB4 (again, whether it is via USB-C connectors or DisplayPort). Anyway, I fail to see the point of combining data & video on a single connector when you are dealing with modern high-res and/or high framerate displays that eat most of the bandwidth.

Well it’s a bit of a chicken and egg problem, isn’t it? How can consumers show a preference for faster storage (or for that matter, higher resolution screens) if they are not offered the option?

Storage is the obvious one, cheap external SSD:s use m.2 connected sticks that in most cases are capable of significantly higher throughput than the USB bottleneck that connect them to the host computer.

Why wouldn’t I, as a consumer, want the full speed of the SSD? The ”only” thing it requires is that the manufacturers migrate to the new standards so that the volume of clients and devices grow. I’m not stupid enough that I can’t see that there is a number of things slowing such migrations down, but lets be honest here - one of those reasons is to make the case for future upgrades, and striving to always have something in your back pocket to drive demand for the next great thing.

Or in some cases, it seems the industry just prefers a stratification of the market, where they can squeeze higher profits out of corporate customers for instance, and having a proliferation of capable consumer devices is seen as a threat (real or imagined) to the bottom line.

It’s difficult to assess from outside the boardrooms and meetings where industry strategy is agreed upon just what the underlying causes are. But I don’t think anyone believes that 5120x2880 screens are five times as expensive to cut than 3840x2160 ones. Or that 10Gbe controllers are almost ten times the price of 1Gbe ones. Or…
Factors other than cost are at play as well.

 

[theluggage]

Storage is the obvious one, cheap external SSD:s use m.2 connected sticks that in most cases are capable of significantly higher throughput than the USB bottleneck that connect them to the host computer.

No, expensive external SSDs use those - cheap external SSDs use SATA 3 connected sticks with a theoretical max of 6Gb/s that is rarely reached vs. 5Gb/s USB 3.1g1 or 10Gb/s USB 3.1g2, while cheap external drives still use spinning rust - still several times cheaper that never gets near that speed. Even those will load your family photos in a second or two and your word documents in a blink of an eye. They're probably OK for preparing your humorous cat videos for Youtube unless you're a gold play button-star who works in 4k. About the only time most general users will get close to the maximum throughput of their drive is when they're running synthetic disc benchmarks.

If you're a 4k video editor, or working with massive Photoshop files and really need that extra speed then Thunderbolt 3 laptops are available and have been for some time - and you'll pay the extra price for those because it's worth your while. Or you'll buy a laptop with an extra internal M.2 slot, or a desktop with PCIe slots, and not need Thunderbolt/USB4.

How can consumers show a preference for faster storage (or for that matter, higher resolution screens) if they are not offered the option?

They've been offered the option for years. Not enough people have bought them to start bringing the prices down, because the majority of customers are happy if their Word docs open in a couple of seconds.

 

[jameskchau]

Steve Jobs beated IBM in 2 crucial battlegrounds: microprocessors and IO ports, nothing else matter more, not even GUI. Jobs has died more than 12 years ago, Cook had barely done enough to make the microprocessor M1 a reality, but Cook fumbled badly in making the Mac Pro (IO Fortress) a reality. True enough TB storage manufacturing is very cost intensive and volatile year-over-year making even the trusted partner LaCie a poor man unable to sell enough Thunderbolt drives to make profits, I still stand with my advice to make Thunderbolt a built in component of the Mac Pro, nothing in the universe can match the usefulness of a full speed Mac Pro with sufficient Thunderbolt 5 storage arrays !!!

 

[Tagbert]

Thunderbolt is an intel tech! that's why it's in PC Boards of higher end! You can even go to Intel site and learn about Thunderbolt is Intel tech!

Thunderbolt was developed by Intel and Apple in a collaboration. Apple has full rights to develop their own implementations and do so in their Apple Silicon chips.

Thunderbolt (interface) - Wikipedia

en.wikipedia.org

 

[MrGunny94]

This is gonna be part of the M3 architecture roadmap for sure. We'll know more with the base M3 if it does come in 2023.

 

[jameskchau]

Thunderbolt greatly improves the ghastly performance lagging because of storage which makes a mockery of all microprocessor advances. I strongly recommend fantastic advancements of the Thunderbolt to bring overall performance far far far the pathetic storage performance we are using now! in my opinion the current storage performances are utterly disgraceful!! Bring in Thunderbolt 10 and beyond, Apple!

 

[Tagbert]

Thunderbolt greatly improves the ghastly performance lagging because of storage which makes a mockery of all microprocessor advances. I strongly recommend fantastic advancements of the Thunderbolt to bring overall performance far far far the pathetic storage performance we are using now! in my opinion the current storage performances are utterly disgraceful!! Bring in Thunderbolt 10 and beyond, Apple!

I’m going to assume that you didn’t use computers back when all we had were spinning disks. If you have you would not be proclaiming “ghastly performance lagging because of storage” on any SSD machine. 😊

 

[diamond.g]

I’m going to assume that you didn’t use computers back when all we had were spinning disks. If you have you would not be proclaiming “ghastly performance lagging because of storage” on any SSD machine. 😊

RAM management definitely mattered a lot back then. You could have a perfectly smooth experience and as soon as you ran out of RAM beachball city!