When that happens, they can officially Take My Money.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Thunderbolt 5 Announced With Up to 3x More Bandwidth
- Thread starter MacRumors
- Start date
- Sort by reaction score
Apple: "iPhone 15 and 15 Pro for sale Friday, in your hands next Friday"
Intel: "The first computer with Thunderbolt 5 shipping sometime in 2024, with supporting accessories likely sometime after 2026"
Yep - M4 at the earliest, if Apple has been working on a USB4 2.0/Thunderbolt 5 hardware implementation in tandem.
M3 was likely taped out months ago.
For those who may not know this, Thunderbolt 4 and 5 are basically conformance marks by Intel. They require support for various hardware and connectivity features, such as USB 3.2, USB4 2.0, and DisplayPort 2.1. For computers, it requires features like MMU protection for the IO.
The 120 Gbps bandwidth and cabling specifications is all USB-IF, specifically USB4 2.0. The poor job the USB-IF does seeking conformance and consistent branding is why you instead see thunderbolt 4 and 5 cable branding.
The 120 Gbps bandwidth and cabling specifications is all USB-IF, specifically USB4 2.0. The poor job the USB-IF does seeking conformance and consistent branding is why you instead see thunderbolt 4 and 5 cable branding.
Errrrr. Mac Studio ... no Intel Processor present.
" ...
Four Thunderbolt 4 ports with support for:
- Thunderbolt 4 (up to 40Gb/s)
- DisplayPort
- USB 4 (up to 40Gb/s)
- USB 3.1 Gen 2 (up to 10Gb/s)
Mac Studio - Technical Specifications
View all the technical specifications for Mac Studio with the M4 Max or M3 Ultra chip.
www.apple.com
That assertion isn't true.
AMD ( and ASMedia) largely just don't want to pay to do the certification test. In part, because they pragmatically don't have to.
It's here! Testing the first AMD laptop with USB4
Razer's new Blade 14, with a 7000-series AMD CPU, is the first Ryzen laptop we've ever seen with full USB4 support right out of the box.
www.pcworld.com
part of USB subculture is doing things the cheapest way you can get away with. But part of this is because AMD isn't putting work into it ( which is cheaper for them.)
Apple 'mentioned; because they don't pass TBv4 certification with their relatively crippled Display out capabilities. The baseline TBv3 can be present, but if you don't cross all the t's and dot all the i's then don't get the TBv4 certification. Part of the process is two 4K display outs. ... And the M1 systems fail.
Later on from when??????? "Sherman set the Wayback machine to November 10, 2020. Sure thing Mr. Peabody.. " ... (from internet archive on the day of the annoucement. )
" Two Thunderbolt / USB 4 ports with support for: "
MacBook Air - Technical Specifications
View the technical specifications for MacBook Air, see what comes in the box, explore configuration options, and get a list of accessories.
web.archive.org
from Macrumors coverage that day...
" Key features of the new MacBook Air:
....
- Two Thunderbolt 3/USB4 ports
"
New MacBook Air Announced as First Apple Silicon Mac With M1 Chip and Fanless Design
Apple today introduced a new MacBook Air as its first Mac with a custom Apple Silicon processor known as the M1 chip. The new MacBook Air is up to...
www.macrumors.com
Apple is not saying "Thunderbolt 4" but they are certainly saying "Thunderbolt".
That's a bit nuts because some devices are Thunderbolt and are peripherals and don't have either Intel or AMD CPUs in them and still get a Thunderbolt label rights.
Vendors who put in the effort to complete TB testing with AMD ( AMD doesn't help has much as they should) have finished this. For example :
" ...
...AMD Socket AM4 for AMD Ryzen™ 5000 Series/ 5000 G-Series/ 4000 G-Series/ 3000 Series/ 3000 G-Series/ 2000 Series/ 2000 G-Series Desktop Processors*...
...
...
2 x Thunderbolt™ 4 port(s) (2 x USB Type-C®) with up to 5V/3A, 15W charging support ..."
https://rog.asus.com/us/motherboards/rog-crosshair/rog-crosshair-viii-extreme-model/spec/
" ... AMD Socket AM4, support for : AMD Ryzen™ 5000 Series/ Ryzen™ 5000 G-Series/ Ryzen™ 4000 G-Series and Ryzen™ 3000 Series Processors
...
Chipset+Intel® Thunderbolt™ 3 Controller:
- 2 x USB Type-C™ ports on the back panel, with USB 3.2 Gen 2 support
B550 VISION D-P (rev. 1.x) Specification | Motherboard - GIGABYTE Global
Lasting Quality from GIGABYTE.GIGABYTE Ultra Durable™ motherboards bring together a unique blend of features and technologies that offer users the absolute ultimate platform fo...
What is lacking on the AMD side is 'one die' or 'in SoC package' support for Thunderbolt . In part, because AMD is mostly outsourcing that to ASMedia and ASMedia is largely covering just base USB4 for those systems vendors who are on a budget.
" ... The ASM2464PD is also designed to be compliant with the Thunderbolt 4 specification and ASMedia is targeting the Thunderbolt certification by the end of Q2 2023. Once the certification has been granted, end-product devices utilizing the ASM2464PD would also be eligible for Thunderbolt certification testing. ..."
https://www.asmedia.com.tw/product/802zX91Yw3tsFgm4/C64ZX59yu4sY1GW5
In 2021-2022 ASMedia USB4 controller was coming "real soon now" . For example,
https://www.techpowerup.com/295030/asmedia-to-launch-usb4-host-controllers-this-year
And yet if look at the description quote from the controllers webpage is it is "oh we will get TBv4 certification end of Q2 2023..."
Intel is setting the certification constraints so that their own laptop SoCs are a very good match. None of Intel's upcoming stuff it going to have a problem with DisplayPort v2.1 because Intel has already shipped DPv2.1 . Three video out streams? Already done a couple generations ago. Full backhaul bandwidth provisioning... again been doing that before Apple every shipped in 2020. etc. etc.
The Thunderbolt certification constraints are clearly set up to make Intel SoCs 'look extremely good'. But they don't overtly block folks who want to put in effort.
There are, and they need Display port(s) connectivity. That kind of DP bandwidth can't be supported via TB3/TB4.
Is Apple tracking USB4v2 and DisplayPort v2.1 ? Yes. Does that means those drive integrated SYSTEM dies timetables more than anything else? No.
Apple integrating the TB controllers into the main die is a dual edged sword. It gets them big wins on Perf/Watt. Costs. etc. But it also couples the controller so billions of other transistors with development schedules that are not the same.
Similarly, TBv5 with approximately zero USB4v2 peripherals really isn't much of a value add. Post 180 above has link to AMD's first USB4 ( the version v4 ) laptop years after USB4 came out. ASMedia still hasn't finished their TBv4 certification ( or just recently finished. ). The vast majority of the PC market is still trying to 'digest' the first version of USB4 let alone this major update (USB4v2 which somewhat understates the changes to a completely new underlying physical signaling protocol. ).
Intel 'previewed' TBv5 over a year ago because USB4v2 is what it is layered on top of. See huge rush of USB4v2 products shipping? Nope. Thunderbolt is not going to evolve as fast as it did when the early stages of TB evolution. Intel and Apple mainly just talking amongst themselves and saying "ship it" was a totally different model. USB-IF is a very large committee with a huge variety of sometimes conflicting vested interest. For better or worse Thunderbolt is deeply coupled to the USB ecosystem. And the USB ecosystem moves way , way , way slower than Intel/Apple did at the early stages. Thunderbolt is going to get very long term support being coupled to the USB ecosystem , but that comes at a cost. Part of that cost is 'update speed'. And another is high control by just 1-2 parties in the standard; what everyone else in the ecosystem matters more to Apple whether they like it or not.
Apple had early access to AV1 YEARS ago. Didn't deploy it 'fixed in stone' on a die until 2023. Intel/AMD CPU packages with PCI-e v5 are deploy currently. Apple PCI-e v5 SoCs in 2023 ... zero. Apple did Afterburner with 'hardware' ProRes far before they did ProRes 'fixed in stone' on an Apple die. Standards where the ink is barely dry on the pragmatic spec are not real good candidate to be thrown onto extremely high volume silicon where they are only a relatively very, very , very small part of the 'value add'. It is far less risky to wait until interoperatibily results are very clear before committing them to being a very small fish in a large pond that would be very expensive to 'fix'.
Nevermind that TBv5 is pretty likely going to 'cost' power to do. More power means lower Pref/watt. Apple is not running toward more lower Pref/Watt ( e.g., PCI-e v5. )
Intel ships a discrete TB controllers. That gives the lots more flexiblity of how fast they can "ship" a controller after the standard settles down. There is going to be a decent gap between Intel discrete and integrated TBv5 controllers. [ pretty good chance that if there was still a desktop Meteor Lake option that Intel was going to couple their discrete controllers to that for a quick rollout. ] Intel isn't rolloing out TBv5 at a super fast past. That is likely exactly why they are talking up "TBv4 being the mainstream focus and running concurrently " because there will be no rapid roll out.
Apple has no discrete controller. And Apple's discrete controller usage is all mainly wrapped up in their very slow moving Display docking stations. ( which do not need DisplayPort v2/2.1 either which is in part why Apple is coming in last place in GPU vendor roll out of that protocol also. )
What the M3's timeframe is now doesn't matter as much as what the M3's timeframe was back when Apple laid out the specs of what was/wasn't going into the design originally. Since TBv5 is pragmatically highly depend upon the USB-IF committee. Betting on when a committee is going to finish is not something you do years in advance of when you think they might finish.
M3's design probably has been primarily done for about a year. (bug fixes , but no feature creep). When the dies ship is a different matter. Chip design is more on the order 3-4 year cycle. Not "oh this snazzy thing appeared last year , let's slap it in" kind of cycles.
Don't really need a GPU to put bandwidth pressure on TBv3. A x4 PCI-e v5 SSD is being impaired also. But also keeps Apple's internal SSD on more competitive footing. Similar issue where Apple is concerned about their own GPU/SSD/etc sales before anything else.
You think it’s about making their SSD look good instead of the physical number of PCIe lanes available?
Apple making more via the mark-up on their SSDs . Yeah I do think they are quite interested in that.
In vast majority of Apple products, there are no substantive PCI-e lanes to hand out anyway. Apple has thrown a high than 'normal' amount of die edge space at the memory subsystem I/O rather than 'waste' it on PCI-e lanes to 3rd party SSD sockets they don't even provision. For a laptop/small-sized enclsoure which is likely limited to just one internal SSD anyway is a far more rational trade-off choice also if trying to maximum overall system performance on a wide variety of common workloads.
Thunderbolt 3 was released Q4 2015, MBP late 2016 already had it. Thunderbolt 4 was released H2 2020, MBP late 2021 already had it. So basically, TB5 was just released, and by the end of the next year, with M3 Pro / Max we should have TB5 ports included in next Macs, IMO.
Last edited:
While the timeline is interesting, I doubt Apple will refresh the design just to include TB5.
Observing how the M3 release schedule appears to be going, TB5 MIGHT align with M3 Ultra release but I suspect TB5 might not be available until M4 Macs are released. 😢
Those were all implemented with discrete Intel chip packages. There is no Apple chip development synchronization there at all. What saying there mostly is that a year after Intel discrete TB controller chips get launched there are new TB systems.
The other issue was that Apple was relatively quite late to USB 3.0. ( there were quirky power control/supplemental PHY chips that had issues early on so there were some good reasons). But Apple was quite slow on the USB 3.0 uptick. If Apple is slow on the USB 4.2 upticks ... that essentially will make it slow on the TB v5 uptick. ( since TB v5 is pragmatically just layered on top of same foundation. Just less optional stuff. )
There aren't any Intel discrete TB controllers in the new systems. The M3 Pro /Max are far, far ,far more coupled to the 'plain' M3 then they are to discrete Intel chips. If M3 is 'too early' for TB then the Pro/Max probably has the same constraints. ( more TB controllers than the smaller plain M3 die got, but the exact same controllers. )
Intel really hasn't released the TB5 discrete controllers yet.
I don't get your scepticism 😉 First of all, Thunderbolt is (or at least was) developed by Intel in collaboration with Apple. I'm more than sure, that Apple has access to all the necessary information way before we knew that TB5 is ready and completed. M3 might have been designed for TB5 since the very beginning, hence the delay. And even if not, I'm pretty sure that a year is enough time for a company like Apple to implement a port interface into their chip. If they're planning to put it into M4 then it would be like 3 years after the release of the interface... Way too late, IMO. Apple has always had top-notch interfaces in their devices (FireWire, TB) and I don't see no reason why would that change...
It was co-developed by Apple and Intel, but Intel owns a great deal of the IP at this point. That means whatever solution Apple comes up with has to either license the IP from Intel, or come up with your own implementation that avoids various Intel patents in addition to meeting a specification that Intel is largely driving. In short, it's an unlikely and uphill battle if Apple wishes to integrate it on the M3.
All of @deconstruct60's points are valid and make perfect sense, but like you I'm still hopeful that Apple can figure something out.
Last edited:
You are talking about Thunderbolt origin history; not where Thunderbolt is now.
Thunderbolt now is an 'add on' layer on top of USB. That's it. It isn't up to Intel or Apple as to how the foudation moves forward. It is up to the whole USB-IF committee. Lots of folks have access to the incremental updates , but they also need to all move together in a larger group ( with various different priorities ).
What Intel is doing now is primarily just mandating certain options that the USB-IF leaves optional ( mostly to gather the consensus needed to get new stuff approved. Factions in USB-IF are cost sensitive and always looking for options to lower bill of material costs ). It is a dual edge sword. it is a bigger collation ( so very little chance of the baseline standard dying over time), but also has higher inertia to change ( so things move slower).
If USB 4v2 is going "no where fast" then so is TB v5. Handwaving at "But Apple invented the name" is mostly misdirection. It has almost nothing to do with how large committee standards move forward over time.
Deeply unlikely. USB 4v2 was under control of the USB-IF committee; not Apple. To boat anchor M3 to a committee that is completely outside Apple's control would be silly. Apple gets to 'merge' with the standard along with everyone else.
A year is too short for a company like Apple ( or any other) for this. Final design stage to chip release is about 6-11 months depending upon all the certifications and validation testing that needs to be done. Throwing money at it doesn't necessarily make that go faster. The design and implementation of M3 was likely wrapped up back in 2022.( that is when N3 was roadmapped to finished up). And the USB4v2 standard didn't complete until end of 2022
USB-C can hit 120Gbps with newly published USB4 Version 2.0 spec
USB-IF’s new USB-C spec supports up to 120Gbps across three lanes.
arstechnica.com
AV1 was finished as a standard multiple years ago (2019) and Apple just got to it (hard baked into silicon).
Care to point out any USB4v2 devices you can buy right now? There aren't any. If release TBv5 before there are any other TBv5 devices what utility are you actually providing. What actual solid validation testing have you actually done? None . ( if there are not finished peripherals devices , you can't possibility have finished testing in a actual production context. )
Apple bought all of those as discrete components. They were not trying to outrun the peripheral's standards adoption rate at any point along those roll outs. Apple bought the same/sibling components that the peripheral folks did. That is how it stays in synch .
Last edited:
The M3 is out and as expected it is Thunderbolt 4.
My question is, will we have to wait for an M4 to have Thunderbolt 5? I assume it is tied directly to the SoC design.
This is the key feature for my next upgrade, either MacMini or MacBook…
USB4 2.0 is the specification that defines how this works; it came out in June. For all we know it may be in the M3 already.
Thunderbolt 5 is Intel's certification mark for USB4 2.0, requiring certain features. This is similar to say AMD FreeSync, which is a certification mark for VESA DisplayPort monitors that have certain feature support.
The Barlow Ridge chip from Intel that supports Thunderbolt 5 comes out next year; I suspect they aren't especially motivated to certify other products before they have something in market.
Technically the M3 is still Thunderbolt 3. ( still stuck with the DispayPort out limitations that preclude TBv4). The M3 Pro/Max .
the 'plain' M4 is very likely still going to be stuck on TBv3 as Apple probably is not going to 'fix' that DisplayController limitation at all. ( Apple designs relatively very large display controllers which is a dual edge sword. The bulk gets them some low power and/or Perf/Watt features want, but on relatively small dies only going to get a fixed amount of controllers on the die; so output stream capped. What makes Apple's controllers bulky is not logic. It is other elements that TSMC N3/N2/etc are not a magical 'get out jail free' card for. )
So don't hold your breath.
Yes, the TB controllers for Apple are directly intergrated inside the main die that Apple designs. That lowers power consumption and raises Perf/Watt. That very likely isn't going to change going forward.
M4 design was probably laid down years ago. So TBv5 probably missed that design feature freeze also. Zero good reason for Apple to weave in a standard that isn't finished onto a die.
It is likely going to be a more expensive Mini Pro or MBP 14"/16" . So save those pennies.
Last edited:
Even more delayed than that though is that the laptop chips (with embedded TB controllers) are not TBv5 ( Meteor Lake , Ultra X 100 series ). That is where the bulk of the momemtum is going to come from. Won't be surprising if it is pretty late in 2024 that see any movement on TBv5. (may see a flurry of demos at CES 2024 about products shipping 'later this year' ).
Apple will release an updated Pro Monitor to go along with new Mac Studio and Mac Pro, that will be 8K, and will be relying on Thunderbolt 5. Both Mac Mini Pro and real pro MPB will be able to run it.