Intel won't have a TB4 PCIe4 controller for at least another year or maybe 2.
In the meantime I think that the first controller actually capable of more than 32 Gb/s would be an ASMedia or Renesas USB4 gen 3x2 PCIe4 x4 host controller. Maybe they will announce such controllers at CES.
Maybe more importantly we should ask when will Intel Intel certify non-Intel platforms, vastly more important than whether PCIe 4.0 would be used given that PCIe 4.0 would not make that much of a difference.
Indeed, this being said the refusal to certify TB on competing platform even after it was proven to be working could be seen as an anti-competitive move. Imagine if AMD just decided to do CPU detection in the driver and refuse to boot if an Intel CPU is detected (ignore the actual market repercussions, just the principle). Would you still find it a reasonable justification that "they want you to buy their products, not a competitor's"? What about if Google said "every Google product and service can only work with Chrome"? Artificially locking out your competitor is the definition of anti-competitive.
Unfortunately (most) news outlets don't want to piss Intel off by calling them out in every relevant article, and US regulators are can barely breath through that tight leash the big companies hold them with to be able to effectively regulate anything.
P.S. Since it actually works I hope AMD or OEMs will bother to integrate it somehow. I don't expect Intel to do the legwork for them, just not actively prevent it.
TB4 is a certification - it certifies that both the TB3 and USB4 are in compliance with the standards and support a minimum level of implementation of those 2 technologies.
One can assume that Intel's TB4 controller would receive TB4 Certification.
@lilkwarrior, indeed there are AMD boards with TB, I own one. But the lack of certification means the controller won't end up integrated in an AMD CPU making it easier for OEMs, and that same missing certification also makes OEMs a bit less than excited about implementing it.
AMD hinted several times that users don't want TB and that may be true at the low end but anyone buying higher end board and CPU combo would definitely go for it.
"Indeed, this being said the refusal to certify TB on competing platform even after it was proven to be working could be seen as an anti-competitive move. Imagine if AMD just decided to do CPU detection in the driver and refuse to boot if an Intel CPU is detected"
1st of all - where is any evidence at all that Intel is refusing to certify TB4 on AMD platforms.
2nd - So AMD is moving in to the OS space - with a competitor to MS Windows - since that would be the only way to keep a system from booting when an Intel CPU is detected - and last I looked AMD has nothing to do with Intel Chipsets for motherboards... And what "driver" would that be again?
@Deicidium369, the evidence is a Google away. It's faster than asking. If you can't get it right I am certainly willing to assist.
Secondly, AMD also makes [size=369]GPUs[/size]* not just CPUs. From a technical standpoint it's perfectly possible to have the GPU driver throw a spanner in the works in any number of ways.
Oh I wish TheinsanegamerN were here to tell me all about those techie ATers who need someone to tell them AMD makes GPUs and they need drivers. All good. Carry on.
*Too bad AT doesn't support any "advanced" features like text formatting...
All I can find is that intel confirmed TB4 works on AMD boards if AMD provides equivalent DMA (which AMD confirmed they do), and that they have certified TB on a few AMD boards already. They also started allowing 3rd parties to make their own TB controllers that have been used on AMD boards. When did they say they will not certify controllers to work on AMD platforms?
Not quite the same thing, SAM is a proprietary technology while TB is ostensibly open to competition and royalty free. Except that Intel retained the certification and so far it's been done with the sole purpose of not certifying any implementation in an AMD CPU. OEMs have some freedom hence my MB can have TB.
For the consumer it would be vastly better to have it actually free, like Intel insists it is in marketing materials. It's a good tech. I'm also certain SAM will be available to Nvidia and Intel in a generation or so.
Thunderbolt 4 is USB4 with mandatory support for Thunderbolt 3 interoperability, as well as various other requirements.
USB4 and Thunderbolt 3 are not the same thing. They use slightly different signaling rates and encoding schemes and also differ in their protocol tunneling capabilities. Unless a USB4 device includes optional support for Thunderbolt 3 Alt Mode, it will not be compatible with Thunderbolt devices. Thunderbolt 3 devices do not support the USB4 protocol.
The PCIe Root Port instances on TGL are only Gen3 (8.0 GT/s) x4. And the configuration is more like 2 dual-port TB4 controllers, each with 2 PCIe Gen3 x4 interfaces.
Regardless, this should allow up to 40 Gbit/s of PCIe tunneling over a single link. We just need someone to benchmark two Thunderbolt SSDs daisy-chained off of a single TGL Thunderbolt 4 port to see that in action.
"... The protocol will also now ride on the new PCIe 4.0 interface, and although that doesn't mean it will be faster, it may serve as the impetus to update the branding. We asked Intel if Thunderbolt 4 is faster than Thunderbolt 3, to which the company responded "more details to come at a later date."
Cool story, but since then Intel has actually released Tiger Lake and the datasheets clearly state that PCIe root ports for the integrated Thunderbolt controllers are Gen3 x4. And the same goes for the ARK listings for the discrete Thunderbolt 4 products.
It's a tech site so accuracy in language is important. For example "VT-d and hyper-threading on an AMD CPU", or "DLSS on Radeon" should raise a flag regardless of technological equivalence. They're not used interchangeably more that you can use "English" and "German" interchangeably because they're both languages that fulfill the same role. Unfortunately most ATers will need to google to see what I mean.
I thought TB 4 part of USB 4, and PCIe 4. I read that in several places. But here it says these controllers work with PCIe 3. What’s going on? I know that Apple said that the M1 has a built-in PCIe 4 bus, so that works.
The USB standards have almost no relationship with the PCIe standards, except that USB 3.0 has reused the physical interface of PCIe 2.0 (PCIe TWO, not 3, i.e. with a 5 Gb/s raw data rate, reduced to 4 Gb/s for the useful data bits).
Also Thunderbolt does not have any direct relationship with PCIe, except that PCIe is one of the communication protocols that can be transported transparently over a Thunderbolt link, so you can connect PCIe peripherals, e.g. external SSDs or external GPUs, over a Thunderbolt cable.
Any Thunderbolt controller has a Thunderbolt interface and a PCIe interface and in theory you could have very different versions on those interfaces, e.g. a Thunderbolt 2 with PCIe 4 or a Thunderbolt 4 with PCIe 1.0.
Obviously most possible combinations do not make sense, so any new Thunderbolt controller will support the most recent Thunderbolt version, i.e. 4, together with a PCIe version allowing maximum compatibility, i.e. PCIe 3, which is also compatible with PCIe4.
Because the Thunderbolt maximum speed for tunneled PCIe is 32 Gb/s, which can be achieved with 4 lanes of PCIe 3, it is unlikely to gain anything by using a PCIe 4 interface, because all consumer CPUs, including the future Intel Rocket Lake, do not have more than 20 PCIe 4 lanes, to be used for 1 GPU + 1 SSD, so all other peripherals, including the Thunderbolt controller, must be connected through PCIe 3 anyway.
You are right, but as you say, that would be mostly for the future, because with current CPUs it is more likely to have 4 free PCIe 3 lanes (usually from the chipset) than to have 2 free PCIe 4 lanes (from the few lanes provided directly by the CPU).
That table (presumably from Intel) is just misleading/wrong/dishonest. The TB3 column should have ticks matching the TB4 one, and as the article points out at the end the bandwidth is many cases was the same. The the difference is a slight tweak to the certification rules requiring 4 PCI, and nothing else. Are Intel actually doing anything properly at the moment?
Like it was explained in the article, what has changed from in the "More bandwidth" row from the table is that now the "32 Gb/s" are guaranteed when you see that "Thunderbolt 4" is claimed.
Until now, you had computers with good Thunderbolt 3 (32 Gb/s) and computers with bad Thunderbolt 3 (16 Gb/s), all of which were sold as computers with Thunderbolt 3.
Unless you could find a review properly benchmarching the Thunderbolt 3, you could usually discover whether a Thunderbolt 3 was good or bad only after buying the product, when it was too late.
The table is also misleading comparing TB3 and USB4 as they have pretty much the same capabilities.
USB 4 gen 3x2 (such a stupid name) is also capable of 40Gb/s, while a certified TB3 controller using 2 PCIe 3.0 lanes is limited to 16Gb/s (and theoretically 4 lanes limit it to 32 Gb/s minus protocol overhead).
Nothing could be farther from the truth. The misleading 'TB3 and USB4 have the same capabilities' concept has become widespread due to two problems:
(a) USB-IF can communicate technical matters to a technical audience perfectly, but not to the average consumer (b) Tech journalists either don't take the trouble to analyze the info provided to them by USB-IF, or plainly want to go in for click-bait headlines like 'USB4 now gives you 40Gbps bandwidth'!
The truth is that the table given in the Intel infographic (that I have reproduced faithfully in the article) is unfortunately true. All the entries in the USB4 column are the ones mandated as minimum for a vendor to get USB4 certification for a particular port in their device. *Some* USB4 implementations can *opt* to do the PCIe tunneling aspects of TB3, since Intel donated those specifications to USB-IF for inclusion in USB4. However, that is not mandatory. So, if you expect all USB4-certified ports to support TB3 devices or TB3 speeds, you are sadly mistaken.
To keep expectations at the minimum, if you see a port advertised as USB4, go in with expectations that it will support USB 3.2 Gen 2 speeds (10Gbps) for data along with DisplayPort tunneling. Any additional features are only gravy and not part of the main course.
USB 3.2 minimum power supply capability is only 4.5W (900mA). Most support 7.5W, but it is not mandatory. That applies to USB 3.2 w/ DP Alternate Mode also.
USB4 makes it mandatory to have a video signal getting tunneled through, and also makes 7.5W power delivery a must.
Most of the confusion about the capabilities stems from mandatory vs. optional to implement components of a spec.
USB4 is a tunneling protocol based on Thunderbolt and supports up to 20 GT/s signaling. That is markedly different than USB 3.2 with DP Alt Mode.
USB 3.2 signaling is only up to 10 GT/s and offers no native provision for tunneling either PCIe or DisplayPort. DP Alt Mode simply replaces the USB signaling on two or all four of the high-speed differential pairs with DisplayPort signaling.
USB4 (and Thunderbolt 4) take USB3, PCIe, and DisplayPort packets and route them over a unified high-speed serial link created by bonding together all four high-speed signaling pairs in a USB Type-C cable. This allows flexible sharing of bandwidth between all supported protocols.
Thank you for quoting a bunch of specifications and entirely missing the point that, to users, if it's not opting to provide any of those apparently optional things (such as PCIe tunnelling), it adds nothing.
I think being able to drive a 4K HDR display without compression while also providing USB bandwidth equivalent to a full USB3 5Gbps link over a single cable is a worthwhile use case. This isn't possible with USB 3.2 and DP Alt Mode, yet with USB4 it only requires 10Gbps (Gen 2) signaling and does not involve any PCIe tunneling.
USB is used in billions of devices annually. Many of those devices are things like phones or Raspberry Pis where the concepts of PCIe tunneling or providing more than 900 mA of bus power make little to no sense. But USB4 also has to scale to desktop workstations where such features are obviously desirable. That's why these features are optional. And when given the choice, most consumers will opt for lower prices and longer battery life rather than simply ticking every box on a list of features they would seldom if ever use.
Including optional features in specifications isn't a bad thing. In the case of USB4, every host will support PCIe tunneling until such time as someone other than Intel or Apple starts shipping an implementation where it isn't included. And if that someone happens to be Qualcomm, Samsung, or HiSilicon, it probably still won't be a problem.
Getting real tired of this "optional" shit with USB & HDMI. It should be mandated that every device must implement every part of the spec in order to receive certification.
That seems preferable, but do you really want cheap electronics to have to choose between an old spec and implementing features they don't need? We don't need universal over speccing, we need clearer communication about capabilities.
That's huge news that the 22Gbps cap has been removed! I've been following TB4/USB4 and this is the first I've heard of that! I'm really pissed that Intel doesn't mention this stuff.
I'm currently booting Linux off an external LaCie Rugged SSD Pro I can move between PC's, which is one of the few Titan Ridge based SSD's that will work via the regular USB-C on my Threadripper box. I really hope we see USB4/TB4 Add-In Cards for AMD systems (that work with any system, not tied to the motherboard).
I just received my new Razer Book 13 (Tiger Lake) which I bought specifically for the dual TB4 ports directly off the CPU. I think it's worth noting that with the Maple Ridge discrete controller, two ports would still share that single PCIe x4 link (and potentially connected to the chipset rather than CPU), whereas the integrated ports on the CPU seem to perform much better.
Is the 22Gbps cap on the host side or the device side or both? Does this mean that my very expensive Titan Ridge SSD is still capped at 22Gbps, and if I want the full 32Gbps I'll need to buy yet another one when Maple Ridge units ship? This was already a replacement for my Samsung X5 which wouldn't work with my AMD box. I also just bought a Helios 3S Titan Ridge enclosure for my SAS card, does this mean I also need to replace that if I want full bandwidth? I hate you Intel.
I don't really understand what "full bandwidth" means to you. 22 Gbps will already get you a PCIe 3 SSD's worth of bandwidth in anything but pure sequential scenarios. Nothing says your external SSD is obsolete.
Then it should blow your mind to learn that there *never was* an arbitrary 22 Gbit/s cap on PCIe throughput.
PCIe tunneling to any single Thunderbolt endpoint is limited to the equivalent of a PCIe Gen3 x4 link. That’s 4 x 8 Gbit/s, or 32 Gbit/s, which is how Intel comes up with their advertised number.
Given an 128 B maximum TLP payload size and 16 B TLP headers, that should yield up to 25.92 Gbit/s of throughput after accounting for encoding and protocol overhead. The observed PCIe throughput over Thunderbolt seems to hit a wall at around 21.32 Gbit/s. However, that would mean that at least 25.92 Gbit/s of PCIe packets are being tunneled... which seems an odd coincidence.
So at best the only improvement in PCIe throughput you could possibly see would be from 21.32 to 25.92 Gbit/s, but I’m pretty sure you won’t even see that because there’s something else at play here.
No, I’m not. That bandwidth isn’t being reserved for anything. The maximum PCIe throughput is limited by encoding, protocol overhead, and implementation details.
Benchmarks clearly show that up to 21.32 Gbit/s PCIe throughput is possible over Thunderbolt. The PCIe and USB4 (which includes everything necessary for Thunderbolt 3 interoperability) specifications are freely available. You can do the math yourself if you like and see that 21.32 Gbit/s of payload data when packetized and framed out for transport over either a PCIe or Thunderbolt link would require at least 25.92 Gbit/s of bandwidth. That infographic from Intel marketing does not account for protocol overhead, and is merely designed to set realistic expectations for maximum PCIe throughput.
There is simply no mechanism anywhere in the specs for limiting PCIe bandwidth to accommodate DisplayPort traffic in the way that everyone seems to insist is happening.
No reoman27, you are wrong. According to the author of this very article, "Thunderbolt 3's bandwidth sharing mechanism between video and data also put in some dampeners – even in the absence of tunneling DisplayPort streams, 18 Gbps of bandwidth was always reserved for video traffic, and only 22 Gbps available for actual data transfer."
Actually, I should clarify that 22 Gbps is the maximum practical data throughput for *TB3*. With TB4, Intel seems to have pushed this up to around 25 Gbps (given their 3000 MBps claim for SSDs over TB4). There is some bandwidth always reserved for video, but Intel never indicated it is *18 Gbps*. I have reached out to Intel for further clarification and may update the article when I get a response. It could just be that up to 10 Gbps is lost in overheads.
Clarification would be great. I don't even know anymore. I read their tech doc (linked above), lots of articles, and I don't know enough to know if the math is right, and it is just overhead, or if there actually is a hardcoded limit, or if it changed, or WTF is going on.
What I do know is, for the amount of generations this has been around, and now with it becoming a standard, and the amount of money I've spent on the tech, that Intel has done a really **itty job at being even the slightest bit forthcoming on what's actually going on here, and what's changing from 3 to 4, etc. Having some BS marketing to sell stuff is one thing, but can we get some ACTUAL ANSWERS somewhere Intel? Anyone?
I'm more interested in TB4 eGPU devices as giving 50% more bandwidth to an eGPU would be a huge win. I currently use an LG Gram 17 + Sonnet Breakaway Box 550 with a 2070 OC installed in it. If I can get a TB4 laptop and TB4 eGPU enclosure, that would be a huge win for setups like mine.
I personally want to see this integrated onto low-end graphics cards for docking purposes (as well as, of course, use with VR headsets). While docking a PC sounds useless, mini-PC form factors are actually pretty portable these days and avoid a lot of the tradeoffs found in laptop form factors.
It'd be ironic if that arrangement also supported an eGPU, but possibly occasionally useful. The benefits are mostly elsewhere, and imo the cost of implementation shouldn't be much higher than the chip itself and I guess some sort of PCIe bridge chip. With PCIe 3, mid-range graphics cards already do not saturate the x16 link, and barely saturate an x8. Power delivery can be increased over the minimums just off of excess from a single 8-pin. And the DisplayPort out is already onboard.
This would primarily benefit the ITX crowd, but I think it would be an extremely sought-after part. There's very few ways to combine an ITX board with Thunderbolt and an AMD CPU.
Thunderbolt 4 is USB4 with Thunderbolt 3 Alternate Mode.
In USB4 signaling mode, Thunderbolt 4 supports two bonded lanes at either 10 GT/s (Gen 2) or 20 GT/s (Gen 3) for a maximum link bandwidth of 40 Gbit/s in both directions at the same time. Existing Thunderbolt 4 controllers support tunneling for up to 10 Gbit/s of USB3 packets over that link.
In USB 3.2 signaling mode, existing Thunderbolt 4 controllers only support single-lane operation at either 5 GT/s (Gen 1) or 10 GT/s (Gen 2) for a maximum link bandwidth of 10 Gbit/s in both directions at the same time. Tunneling of other protocols is not supported.
A separate USB 2.0 bus is also included, supporting a single half-duplex channel at 1.5 Mbit/s (low-speed), 12 Mbit/s (full-speed), and 480 Mbit/s (high-speed) for a maximum link bandwidth of 480 Mbit/s in one direction at a time.
Intel didn't comment on one crucial part of TB4, is the Intel VT-d a crucial part of the spec. or TB4 can work with other IO-MMU Virtualisation standards as long as they meet the minimum requirements set by intel.
The crucial about this is that in the first case, no body beside intel will be able to provide TB4 on their platform unless intel also license them to have Intel VT-d. In the second case, Both AMD and Apple could also support TB4 later using their own Virtualisation technologies.Like AMD using their own AMD-Vi (their own branding for IO-MMU). I don't know about Apple in this regard.
But I guess, both this and PCIe 4.0 on TB will need more time, it's rumoured that AMD won't have USB 4.0 except in 2022. Though motherboard makers could add it sooner if other chipset makers made it sooner enough before AMD have it native.
Rene Ritchie supposedly received confirmation from Intel back in July that Thunderbolt 4 certification will be available for machines that do not feature Intel silicon / VT-d. https://www.imore.com/apple-silicon-macs-will-cont...
Making Thunderbolt controllers with a PCIe Gen4 interface is probably relatively easy for Intel. Sorting out their process woes and what that has done to their current client CPU offerings is a bit more challenging. Once Intel starts to offer a client PCH with PCIe Gen4, I'm sure we'll see the Thunderbolt controllers follow suit.
Sadly, the controller is only a small part of the total costs, that's why some TB versions of the exact same device can have $50 premium, but mostly can reach $100 premium.
That's Mouser's pricing, not Intel's customer pricing. Intel's pricing for dual-port Thunderbolt 3 controllers was 8.55 - $9.65. The quad-port Goshen Ridge JHL8440 is listed as $10.15, and I'd expect the JHL8540 to slot in under that.
But either way, you're saying a $1.50 increase to the BOM makes Thunderbolt peripherals DOA now that Intel is shipping millions of integrated host ports at the same time?
When did companies fall back to reporting bandwidth numbers as half duplex x2? You don't assume 1GbE is 500Mbps each direction, and "1Gb" only when you add them together.
I wonder if that's the silicon behind the TB4 dock OWC has been offering for pre-order for months, while the shipping date keeps slipping...
With TB3 backward connectivity I have been ruminating about that to network some of my NUCs into a 40Gbit/s cluster on the cheap and small, using TCP/IP over TB.
That would be the Goshen Ridge JHL8440 in the OWC Thunderbolt Hub / Dock. The folks that put in for the initial pre-order are just starting to receive units now, according to reports over at MacRumors.
More looking at Sonnet's TB3 to SFP+ for the upcoming NUC11 Pro models - been testing on my Dell XPS 13 with the 1165G7 - and full 10Gb/s line speed is easily achieved and sustained - and I can finally use the fiber I ran throughout - and will be able to use the full 10Gb/s port speed on the Cisco switches rather than burning a port with a 2.5Gb/s copper connection.
I really love and appreciate what Intel has done with Thunderbolt from the initial development to the latest Thunderbolt 4. With Thunderbolt 4 it is clear to see Intel's commitment to making sure Thunderbolt is really the best USB-C implementation.
Thunderbolt 4 is really great on the device side with the Goshen Ridge JHL8440 Thunderbolt 4 controller because it combines the best feature from USB, the new Multi-Port Accessory Architecture with the Daisy Chain Accessory Architecture in one product. Also having full bandwidth when a Thunderbolt 4 devices is connected to a USB4 40 host with Thunderbolt 3 compatibility which gives a 360 degrees full circle relationship.
My ideal Thunderbolt 4 dock includes 4x Thunderbolt 4/USB4 USB-C ports (1UP+3DWN), 2 Display (HDMI 2.0 and/or DisplayPort 1.4), 4x USB 3.2 USB-A with quick charge, 1x 10GbE with AVB support.
Moving forward Intel will likely have Thunderbolt integrated to desktop CPUs just as they did with the mobile CPUs like Ice Lake and Tiger Lake. However it will be great to see AMD motherboards with a Maple Ridge Controller to add Thunderbolt 4 and USB4 support. It will be great to see ARM support thunderbolt on something like the Surface Pro X. It will be even better if Microsoft ditched their surface connect and just used Thunderbolt 4 on their next Surface Computers.
I can't wait for Intel to move Thunderbolt to PCIe 4.0 and DisplayPort 2.0; PCIe 3.0 and DisplayPort 1.4 are already bottlenecks.
Many 500 series Intel motherboards will feature the Maple Ridge. Outside Apple's M1 Thunderbolt integration, there have been no other 3rd party Thunderbolt controllers announced. Gigabyte have historically led the way for Intel motherboards with Thunderbolt support. AsRock on the other hand took the lead with support for Thunderbolt on their AMD motherboards. Alder Lake will most probably be the 1st generation to feature Thunderbolt integrated to both Desktop and Laptop CPUs. I'm just hoping it will also means upgrading to PCIe 4.0.
We can expect every Thunderbolt 4 Host to act the same regardless of platform. We can expect Thunderbolt 4 devices to have better cross platform performance. We can expect the Thunderbolt 4 cable to work regardless of protocol or platform. It really is unfortunate that the first ARM Based Apple M1 Silicon doesn't get Thunderbolt 4 certification mainly due to only supporting one external display and maybe no eGPU support.
The is only one thing Thunderbolt 4 doesn't fix is BANDWIDTH. Thunderbolt 4 is still PCIe 3.0 x4 and 2x DisplayPort 1.4, which is already a bottleneck with PCIe 4.0 SSDs and 8K HDR 60Hz Display Panels.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
79 Comments
Back to Article
dwillmore - Wednesday, December 23, 2020 - link
Is there a roadmap for them to support PCI-E 4.0 like modern processors use?Ryan Smith - Wednesday, December 23, 2020 - link
Tiger Lake in mobile, Rocket Lake on the desktop.dwillmore - Wednesday, December 23, 2020 - link
Those are processors, not TB controllers. Is there a roadmap for having TB controllers that support PCI-E 4?Arsenica - Wednesday, December 23, 2020 - link
Intel won't have a TB4 PCIe4 controller for at least another year or maybe 2.In the meantime I think that the first controller actually capable of more than 32 Gb/s would be an ASMedia or Renesas USB4 gen 3x2 PCIe4 x4 host controller. Maybe they will announce such controllers at CES.
at_clucks - Thursday, December 24, 2020 - link
Maybe more importantly we should ask when will Intel Intel certify non-Intel platforms, vastly more important than whether PCIe 4.0 would be used given that PCIe 4.0 would not make that much of a difference.DigitalFreak - Thursday, December 24, 2020 - link
They may never do that. After all, they want you to buy and Intel CPU, not a competitor's.at_clucks - Saturday, December 26, 2020 - link
Indeed, this being said the refusal to certify TB on competing platform even after it was proven to be working could be seen as an anti-competitive move. Imagine if AMD just decided to do CPU detection in the driver and refuse to boot if an Intel CPU is detected (ignore the actual market repercussions, just the principle). Would you still find it a reasonable justification that "they want you to buy their products, not a competitor's"? What about if Google said "every Google product and service can only work with Chrome"? Artificially locking out your competitor is the definition of anti-competitive.Unfortunately (most) news outlets don't want to piss Intel off by calling them out in every relevant article, and US regulators are can barely breath through that tight leash the big companies hold them with to be able to effectively regulate anything.
at_clucks - Saturday, December 26, 2020 - link
P.S. Since it actually works I hope AMD or OEMs will bother to integrate it somehow. I don't expect Intel to do the legwork for them, just not actively prevent it.lilkwarrior - Saturday, December 26, 2020 - link
The are AMD Thunderbolt 3 Motherboards already; they will be Thunderbolt 4 AMD Motherboards either via USB4 or dedicated Thunderbolt 4 controllers.lilkwarrior - Saturday, December 26, 2020 - link
*USB4 is more of a guarantee of Thunderbolt 3 capabilities being more commonplace in AMD Motherboards moving forwardDeicidium369 - Sunday, December 27, 2020 - link
TB4 is a certification - it certifies that both the TB3 and USB4 are in compliance with the standards and support a minimum level of implementation of those 2 technologies.One can assume that Intel's TB4 controller would receive TB4 Certification.
at_clucks - Sunday, December 27, 2020 - link
@lilkwarrior, indeed there are AMD boards with TB, I own one. But the lack of certification means the controller won't end up integrated in an AMD CPU making it easier for OEMs, and that same missing certification also makes OEMs a bit less than excited about implementing it.AMD hinted several times that users don't want TB and that may be true at the low end but anyone buying higher end board and CPU combo would definitely go for it.
Deicidium369 - Sunday, December 27, 2020 - link
"Indeed, this being said the refusal to certify TB on competing platform even after it was proven to be working could be seen as an anti-competitive move. Imagine if AMD just decided to do CPU detection in the driver and refuse to boot if an Intel CPU is detected"1st of all - where is any evidence at all that Intel is refusing to certify TB4 on AMD platforms.
2nd - So AMD is moving in to the OS space - with a competitor to MS Windows - since that would be the only way to keep a system from booting when an Intel CPU is detected - and last I looked AMD has nothing to do with Intel Chipsets for motherboards... And what "driver" would that be again?
at_clucks - Sunday, December 27, 2020 - link
@Deicidium369, the evidence is a Google away. It's faster than asking. If you can't get it right I am certainly willing to assist.Secondly, AMD also makes [size=369]GPUs[/size]* not just CPUs. From a technical standpoint it's perfectly possible to have the GPU driver throw a spanner in the works in any number of ways.
Oh I wish TheinsanegamerN were here to tell me all about those techie ATers who need someone to tell them AMD makes GPUs and they need drivers. All good. Carry on.
*Too bad AT doesn't support any "advanced" features like text formatting...
whatthe123 - Sunday, December 27, 2020 - link
All I can find is that intel confirmed TB4 works on AMD boards if AMD provides equivalent DMA (which AMD confirmed they do), and that they have certified TB on a few AMD boards already. They also started allowing 3rd parties to make their own TB controllers that have been used on AMD boards. When did they say they will not certify controllers to work on AMD platforms?http://www.asrock.com/news/index.asp?iD=4434
phatboye - Sunday, December 27, 2020 - link
Pretty sure that AMD Radeon 6000 series currently do not support SAM on Intel platforms.at_clucks - Tuesday, December 29, 2020 - link
Not quite the same thing, SAM is a proprietary technology while TB is ostensibly open to competition and royalty free. Except that Intel retained the certification and so far it's been done with the sole purpose of not certifying any implementation in an AMD CPU. OEMs have some freedom hence my MB can have TB.For the consumer it would be vastly better to have it actually free, like Intel insists it is in marketing materials. It's a good tech. I'm also certain SAM will be available to Nvidia and Intel in a generation or so.
headeffects - Thursday, December 24, 2020 - link
That would be a new version of thunderbolt, ie TB5. 4 is just rolling out.LiKenun - Friday, December 25, 2020 - link
Then they'd better quadruple the current speeds since they didn't double the speed as with previous generations.Deicidium369 - Sunday, December 27, 2020 - link
TB4 is the same speed as TB3 - TB4 is a Certification.repoman27 - Sunday, December 27, 2020 - link
Thunderbolt 4 is USB4 with mandatory support for Thunderbolt 3 interoperability, as well as various other requirements.USB4 and Thunderbolt 3 are not the same thing. They use slightly different signaling rates and encoding schemes and also differ in their protocol tunneling capabilities. Unless a USB4 device includes optional support for Thunderbolt 3 Alt Mode, it will not be compatible with Thunderbolt devices. Thunderbolt 3 devices do not support the USB4 protocol.
kwinz - Monday, February 1, 2021 - link
Also I understood that Thunderbolt Alternate Mode support for USB4 HUBs is mandatory - even without Thunderbolt 4 certification.JayNor - Wednesday, December 23, 2020 - link
Is there a roadmap for AMD to support pcie4 on APUs like modern TGL APUs provide?dwillmore - Wednesday, December 23, 2020 - link
Tigerlake isn't an APU, nor it is a TB controller.Deicidium369 - Wednesday, December 23, 2020 - link
APU is an AMD name for a CPU+GPU. by every metric Tiger Lake is an APU.jeremyshaw - Wednesday, December 23, 2020 - link
Tiger Lake has 4 TB4 controllers integrated into the main IC, all connected internally via PCIe 4.0.repoman27 - Thursday, December 24, 2020 - link
The PCIe Root Port instances on TGL are only Gen3 (8.0 GT/s) x4. And the configuration is more like 2 dual-port TB4 controllers, each with 2 PCIe Gen3 x4 interfaces.Regardless, this should allow up to 40 Gbit/s of PCIe tunneling over a single link. We just need someone to benchmark two Thunderbolt SSDs daisy-chained off of a single TGL Thunderbolt 4 port to see that in action.
Deicidium369 - Sunday, December 27, 2020 - link
"... The protocol will also now ride on the new PCIe 4.0 interface, and although that doesn't mean it will be faster, it may serve as the impetus to update the branding. We asked Intel if Thunderbolt 4 is faster than Thunderbolt 3, to which the company responded "more details to come at a later date."https://www.tomshardware.com/news/what-is-thunderb...
repoman27 - Sunday, December 27, 2020 - link
Cool story, but since then Intel has actually released Tiger Lake and the datasheets clearly state that PCIe root ports for the integrated Thunderbolt controllers are Gen3 x4. And the same goes for the ARK listings for the discrete Thunderbolt 4 products.repoman27 - Thursday, December 24, 2020 - link
To clarify, that's just the PCIe Root Port instances for the integrated Thunderbolt 4 controllers on TGL that are Gen3.dotjaz - Thursday, December 24, 2020 - link
Then what is it in AMD terms? It has everything APU has, and then some.at_clucks - Saturday, December 26, 2020 - link
It's a tech site so accuracy in language is important. For example "VT-d and hyper-threading on an AMD CPU", or "DLSS on Radeon" should raise a flag regardless of technological equivalence. They're not used interchangeably more that you can use "English" and "German" interchangeably because they're both languages that fulfill the same role. Unfortunately most ATers will need to google to see what I mean.melgross - Wednesday, December 23, 2020 - link
I thought TB 4 part of USB 4, and PCIe 4. I read that in several places. But here it says these controllers work with PCIe 3. What’s going on? I know that Apple said that the M1 has a built-in PCIe 4 bus, so that works.AdrianBc - Wednesday, December 23, 2020 - link
The USB standards have almost no relationship with the PCIe standards, except that USB 3.0 has reused the physical interface of PCIe 2.0 (PCIe TWO, not 3, i.e. with a 5 Gb/s raw data rate, reduced to 4 Gb/s for the useful data bits).Also Thunderbolt does not have any direct relationship with PCIe, except that PCIe is one of the communication protocols that can be transported transparently over a Thunderbolt link, so you can connect PCIe peripherals, e.g. external SSDs or external GPUs, over a Thunderbolt cable.
Any Thunderbolt controller has a Thunderbolt interface and a PCIe interface and in theory you could have very different versions on those interfaces, e.g. a Thunderbolt 2 with PCIe 4 or a Thunderbolt 4 with PCIe 1.0.
Obviously most possible combinations do not make sense, so any new Thunderbolt controller will support the most recent Thunderbolt version, i.e. 4, together with a PCIe version allowing maximum compatibility, i.e. PCIe 3, which is also compatible with PCIe4.
Because the Thunderbolt maximum speed for tunneled PCIe is 32 Gb/s, which can be achieved with 4 lanes of PCIe 3, it is unlikely to gain anything by using a PCIe 4 interface, because all consumer CPUs, including the future Intel Rocket Lake, do not have more than 20 PCIe 4 lanes, to be used for 1 GPU + 1 SSD, so all other peripherals, including the Thunderbolt controller, must be connected through PCIe 3 anyway.
invinciblegod - Wednesday, December 23, 2020 - link
Wouldn't the benefit of 4.0 be that instead of an x4 link, in the future, Thunderbolt 4 can use an x2 link instead and save on pcie lanes?AdrianBc - Wednesday, December 23, 2020 - link
You are right, but as you say, that would be mostly for the future, because with current CPUs it is more likely to have 4 free PCIe 3 lanes (usually from the chipset) than to have 2 free PCIe 4 lanes (from the few lanes provided directly by the CPU).dontlistentome - Wednesday, December 23, 2020 - link
That table (presumably from Intel) is just misleading/wrong/dishonest.The TB3 column should have ticks matching the TB4 one, and as the article points out at the end the bandwidth is many cases was the same.
The the difference is a slight tweak to the certification rules requiring 4 PCI, and nothing else.
Are Intel actually doing anything properly at the moment?
AdrianBc - Wednesday, December 23, 2020 - link
Like it was explained in the article, what has changed from in the "More bandwidth" row from the table is that now the "32 Gb/s" are guaranteed when you see that "Thunderbolt 4" is claimed.Until now, you had computers with good Thunderbolt 3 (32 Gb/s) and computers with bad Thunderbolt 3 (16 Gb/s), all of which were sold as computers with Thunderbolt 3.
Unless you could find a review properly benchmarching the Thunderbolt 3, you could usually discover whether a Thunderbolt 3 was good or bad only after buying the product, when it was too late.
At least Intel has corrected this issue now.
Arsenica - Wednesday, December 23, 2020 - link
The table is also misleading comparing TB3 and USB4 as they have pretty much the same capabilities.USB 4 gen 3x2 (such a stupid name) is also capable of 40Gb/s, while a certified TB3 controller using 2 PCIe 3.0 lanes is limited to 16Gb/s (and theoretically 4 lanes limit it to 32 Gb/s minus protocol overhead).
ganeshts - Wednesday, December 23, 2020 - link
Nothing could be farther from the truth. The misleading 'TB3 and USB4 have the same capabilities' concept has become widespread due to two problems:(a) USB-IF can communicate technical matters to a technical audience perfectly, but not to the average consumer
(b) Tech journalists either don't take the trouble to analyze the info provided to them by USB-IF, or plainly want to go in for click-bait headlines like 'USB4 now gives you 40Gbps bandwidth'!
The truth is that the table given in the Intel infographic (that I have reproduced faithfully in the article) is unfortunately true. All the entries in the USB4 column are the ones mandated as minimum for a vendor to get USB4 certification for a particular port in their device. *Some* USB4 implementations can *opt* to do the PCIe tunneling aspects of TB3, since Intel donated those specifications to USB-IF for inclusion in USB4. However, that is not mandatory. So, if you expect all USB4-certified ports to support TB3 devices or TB3 speeds, you are sadly mistaken.
To keep expectations at the minimum, if you see a port advertised as USB4, go in with expectations that it will support USB 3.2 Gen 2 speeds (10Gbps) for data along with DisplayPort tunneling. Any additional features are only gravy and not part of the main course.
hubick - Wednesday, December 23, 2020 - link
Which adds basically nothing to regular USB 3.2 w DP Alternate Mode.ganeshts - Wednesday, December 23, 2020 - link
USB 3.2 minimum power supply capability is only 4.5W (900mA). Most support 7.5W, but it is not mandatory. That applies to USB 3.2 w/ DP Alternate Mode also.USB4 makes it mandatory to have a video signal getting tunneled through, and also makes 7.5W power delivery a must.
Most of the confusion about the capabilities stems from mandatory vs. optional to implement components of a spec.
repoman27 - Wednesday, December 23, 2020 - link
USB4 is a tunneling protocol based on Thunderbolt and supports up to 20 GT/s signaling. That is markedly different than USB 3.2 with DP Alt Mode.USB 3.2 signaling is only up to 10 GT/s and offers no native provision for tunneling either PCIe or DisplayPort. DP Alt Mode simply replaces the USB signaling on two or all four of the high-speed differential pairs with DisplayPort signaling.
USB4 (and Thunderbolt 4) take USB3, PCIe, and DisplayPort packets and route them over a unified high-speed serial link created by bonding together all four high-speed signaling pairs in a USB Type-C cable. This allows flexible sharing of bandwidth between all supported protocols.
hubick - Thursday, December 24, 2020 - link
Thank you for quoting a bunch of specifications and entirely missing the point that, to users, if it's not opting to provide any of those apparently optional things (such as PCIe tunnelling), it adds nothing.timecop1818 - Thursday, December 24, 2020 - link
You're an idiot, repoman27 knows more about USB and DisplayPort than you ever will.repoman27 - Thursday, December 24, 2020 - link
I think being able to drive a 4K HDR display without compression while also providing USB bandwidth equivalent to a full USB3 5Gbps link over a single cable is a worthwhile use case. This isn't possible with USB 3.2 and DP Alt Mode, yet with USB4 it only requires 10Gbps (Gen 2) signaling and does not involve any PCIe tunneling.USB is used in billions of devices annually. Many of those devices are things like phones or Raspberry Pis where the concepts of PCIe tunneling or providing more than 900 mA of bus power make little to no sense. But USB4 also has to scale to desktop workstations where such features are obviously desirable. That's why these features are optional. And when given the choice, most consumers will opt for lower prices and longer battery life rather than simply ticking every box on a list of features they would seldom if ever use.
Including optional features in specifications isn't a bad thing. In the case of USB4, every host will support PCIe tunneling until such time as someone other than Intel or Apple starts shipping an implementation where it isn't included. And if that someone happens to be Qualcomm, Samsung, or HiSilicon, it probably still won't be a problem.
DigitalFreak - Thursday, December 24, 2020 - link
Getting real tired of this "optional" shit with USB & HDMI. It should be mandated that every device must implement every part of the spec in order to receive certification.dipique - Thursday, December 24, 2020 - link
That seems preferable, but do you really want cheap electronics to have to choose between an old spec and implementing features they don't need? We don't need universal over speccing, we need clearer communication about capabilities.lilkwarrior - Saturday, December 26, 2020 - link
Removal of Thunderbolt from USB4 is primary going to be a mobile device thing, not the defaulthubick - Wednesday, December 23, 2020 - link
That's huge news that the 22Gbps cap has been removed! I've been following TB4/USB4 and this is the first I've heard of that! I'm really pissed that Intel doesn't mention this stuff.I'm currently booting Linux off an external LaCie Rugged SSD Pro I can move between PC's, which is one of the few Titan Ridge based SSD's that will work via the regular USB-C on my Threadripper box. I really hope we see USB4/TB4 Add-In Cards for AMD systems (that work with any system, not tied to the motherboard).
I just received my new Razer Book 13 (Tiger Lake) which I bought specifically for the dual TB4 ports directly off the CPU. I think it's worth noting that with the Maple Ridge discrete controller, two ports would still share that single PCIe x4 link (and potentially connected to the chipset rather than CPU), whereas the integrated ports on the CPU seem to perform much better.
Is the 22Gbps cap on the host side or the device side or both? Does this mean that my very expensive Titan Ridge SSD is still capped at 22Gbps, and if I want the full 32Gbps I'll need to buy yet another one when Maple Ridge units ship? This was already a replacement for my Samsung X5 which wouldn't work with my AMD box. I also just bought a Helios 3S Titan Ridge enclosure for my SAS card, does this mean I also need to replace that if I want full bandwidth? I hate you Intel.
lmcd - Wednesday, December 23, 2020 - link
I don't really understand what "full bandwidth" means to you. 22 Gbps will already get you a PCIe 3 SSD's worth of bandwidth in anything but pure sequential scenarios. Nothing says your external SSD is obsolete.repoman27 - Wednesday, December 23, 2020 - link
Then it should blow your mind to learn that there *never was* an arbitrary 22 Gbit/s cap on PCIe throughput.PCIe tunneling to any single Thunderbolt endpoint is limited to the equivalent of a PCIe Gen3 x4 link. That’s 4 x 8 Gbit/s, or 32 Gbit/s, which is how Intel comes up with their advertised number.
Given an 128 B maximum TLP payload size and 16 B TLP headers, that should yield up to 25.92 Gbit/s of throughput after accounting for encoding and protocol overhead. The observed PCIe throughput over Thunderbolt seems to hit a wall at around 21.32 Gbit/s. However, that would mean that at least 25.92 Gbit/s of PCIe packets are being tunneled... which seems an odd coincidence.
So at best the only improvement in PCIe throughput you could possibly see would be from 21.32 to 25.92 Gbit/s, but I’m pretty sure you won’t even see that because there’s something else at play here.
hubick - Wednesday, December 23, 2020 - link
You are incorrect. Thunderbolt 3 reserves bandwidth for DisplayPort, even when it's not in use.See Figure 7. https://thunderbolttechnology.net/sites/default/fi...
The maximum bandwidth for data, when only data is being transmit, is 22Gbps.
repoman27 - Wednesday, December 23, 2020 - link
No, I’m not. That bandwidth isn’t being reserved for anything. The maximum PCIe throughput is limited by encoding, protocol overhead, and implementation details.Benchmarks clearly show that up to 21.32 Gbit/s PCIe throughput is possible over Thunderbolt. The PCIe and USB4 (which includes everything necessary for Thunderbolt 3 interoperability) specifications are freely available. You can do the math yourself if you like and see that 21.32 Gbit/s of payload data when packetized and framed out for transport over either a PCIe or Thunderbolt link would require at least 25.92 Gbit/s of bandwidth. That infographic from Intel marketing does not account for protocol overhead, and is merely designed to set realistic expectations for maximum PCIe throughput.
There is simply no mechanism anywhere in the specs for limiting PCIe bandwidth to accommodate DisplayPort traffic in the way that everyone seems to insist is happening.
vlad42 - Friday, December 25, 2020 - link
No reoman27, you are wrong. According to the author of this very article, "Thunderbolt 3's bandwidth sharing mechanism between video and data also put in some dampeners – even in the absence of tunneling DisplayPort streams, 18 Gbps of bandwidth was always reserved for video traffic, and only 22 Gbps available for actual data transfer."ganeshts - Saturday, December 26, 2020 - link
Actually, I should clarify that 22 Gbps is the maximum practical data throughput for *TB3*. With TB4, Intel seems to have pushed this up to around 25 Gbps (given their 3000 MBps claim for SSDs over TB4). There is some bandwidth always reserved for video, but Intel never indicated it is *18 Gbps*. I have reached out to Intel for further clarification and may update the article when I get a response. It could just be that up to 10 Gbps is lost in overheads.hubick - Tuesday, December 29, 2020 - link
Clarification would be great. I don't even know anymore. I read their tech doc (linked above), lots of articles, and I don't know enough to know if the math is right, and it is just overhead, or if there actually is a hardcoded limit, or if it changed, or WTF is going on.What I do know is, for the amount of generations this has been around, and now with it becoming a standard, and the amount of money I've spent on the tech, that Intel has done a really **itty job at being even the slightest bit forthcoming on what's actually going on here, and what's changing from 3 to 4, etc. Having some BS marketing to sell stuff is one thing, but can we get some ACTUAL ANSWERS somewhere Intel? Anyone?
hfm - Friday, December 25, 2020 - link
I'm more interested in TB4 eGPU devices as giving 50% more bandwidth to an eGPU would be a huge win. I currently use an LG Gram 17 + Sonnet Breakaway Box 550 with a 2070 OC installed in it. If I can get a TB4 laptop and TB4 eGPU enclosure, that would be a huge win for setups like mine.lmcd - Wednesday, December 23, 2020 - link
I personally want to see this integrated onto low-end graphics cards for docking purposes (as well as, of course, use with VR headsets). While docking a PC sounds useless, mini-PC form factors are actually pretty portable these days and avoid a lot of the tradeoffs found in laptop form factors.It'd be ironic if that arrangement also supported an eGPU, but possibly occasionally useful. The benefits are mostly elsewhere, and imo the cost of implementation shouldn't be much higher than the chip itself and I guess some sort of PCIe bridge chip. With PCIe 3, mid-range graphics cards already do not saturate the x16 link, and barely saturate an x8. Power delivery can be increased over the minimums just off of excess from a single 8-pin. And the DisplayPort out is already onboard.
This would primarily benefit the ITX crowd, but I think it would be an extremely sought-after part. There's very few ways to combine an ITX board with Thunderbolt and an AMD CPU.
headeffects - Thursday, December 24, 2020 - link
What’s the actual USB spec for their 4.0 implementation. Is it 20gbps?repoman27 - Thursday, December 24, 2020 - link
Thunderbolt 4 is USB4 with Thunderbolt 3 Alternate Mode.In USB4 signaling mode, Thunderbolt 4 supports two bonded lanes at either 10 GT/s (Gen 2) or 20 GT/s (Gen 3) for a maximum link bandwidth of 40 Gbit/s in both directions at the same time. Existing Thunderbolt 4 controllers support tunneling for up to 10 Gbit/s of USB3 packets over that link.
In USB 3.2 signaling mode, existing Thunderbolt 4 controllers only support single-lane operation at either 5 GT/s (Gen 1) or 10 GT/s (Gen 2) for a maximum link bandwidth of 10 Gbit/s in both directions at the same time. Tunneling of other protocols is not supported.
A separate USB 2.0 bus is also included, supporting a single half-duplex channel at 1.5 Mbit/s (low-speed), 12 Mbit/s (full-speed), and 480 Mbit/s (high-speed) for a maximum link bandwidth of 480 Mbit/s in one direction at a time.
Xajel - Thursday, December 24, 2020 - link
Intel didn't comment on one crucial part of TB4, is the Intel VT-d a crucial part of the spec. or TB4 can work with other IO-MMU Virtualisation standards as long as they meet the minimum requirements set by intel.The crucial about this is that in the first case, no body beside intel will be able to provide TB4 on their platform unless intel also license them to have Intel VT-d.
In the second case, Both AMD and Apple could also support TB4 later using their own Virtualisation technologies.Like AMD using their own AMD-Vi (their own branding for IO-MMU). I don't know about Apple in this regard.
But I guess, both this and PCIe 4.0 on TB will need more time, it's rumoured that AMD won't have USB 4.0 except in 2022. Though motherboard makers could add it sooner if other chipset makers made it sooner enough before AMD have it native.
repoman27 - Thursday, December 24, 2020 - link
Rene Ritchie supposedly received confirmation from Intel back in July that Thunderbolt 4 certification will be available for machines that do not feature Intel silicon / VT-d. https://www.imore.com/apple-silicon-macs-will-cont...Making Thunderbolt controllers with a PCIe Gen4 interface is probably relatively easy for Intel. Sorting out their process woes and what that has done to their current client CPU offerings is a bit more challenging. Once Intel starts to offer a client PCH with PCIe Gen4, I'm sure we'll see the Thunderbolt controllers follow suit.
The_Assimilator - Thursday, December 24, 2020 - link
$11 per controller makes TB4 DOA, regardless of how many ports Intel wastes silicon on integrating it into TGL.James5mith - Thursday, December 24, 2020 - link
I would happily pay an $11 premium for an external SSD enclosure that uses TB4/USB4 dual-mode.Xajel - Sunday, December 27, 2020 - link
Sadly, the controller is only a small part of the total costs, that's why some TB versions of the exact same device can have $50 premium, but mostly can reach $100 premium.repoman27 - Thursday, December 24, 2020 - link
That's Mouser's pricing, not Intel's customer pricing. Intel's pricing for dual-port Thunderbolt 3 controllers was 8.55 - $9.65. The quad-port Goshen Ridge JHL8440 is listed as $10.15, and I'd expect the JHL8540 to slot in under that.But either way, you're saying a $1.50 increase to the BOM makes Thunderbolt peripherals DOA now that Intel is shipping millions of integrated host ports at the same time?
James5mith - Thursday, December 24, 2020 - link
When did companies fall back to reporting bandwidth numbers as half duplex x2? You don't assume 1GbE is 500Mbps each direction, and "1Gb" only when you add them together.Why is the industry being dumb?
repoman27 - Thursday, December 24, 2020 - link
Where is that happening here?abufrejoval - Thursday, December 24, 2020 - link
I wonder if that's the silicon behind the TB4 dock OWC has been offering for pre-order for months, while the shipping date keeps slipping...With TB3 backward connectivity I have been ruminating about that to network some of my NUCs into a 40Gbit/s cluster on the cheap and small, using TCP/IP over TB.
repoman27 - Thursday, December 24, 2020 - link
That would be the Goshen Ridge JHL8440 in the OWC Thunderbolt Hub / Dock. The folks that put in for the initial pre-order are just starting to receive units now, according to reports over at MacRumors.Deicidium369 - Sunday, December 27, 2020 - link
More looking at Sonnet's TB3 to SFP+ for the upcoming NUC11 Pro models - been testing on my Dell XPS 13 with the 1165G7 - and full 10Gb/s line speed is easily achieved and sustained - and I can finally use the fiber I ran throughout - and will be able to use the full 10Gb/s port speed on the Cisco switches rather than burning a port with a 2.5Gb/s copper connection.KimGitz - Saturday, December 26, 2020 - link
I really love and appreciate what Intel has done with Thunderbolt from the initial development to the latest Thunderbolt 4. With Thunderbolt 4 it is clear to see Intel's commitment to making sure Thunderbolt is really the best USB-C implementation.Thunderbolt 4 is really great on the device side with the Goshen Ridge JHL8440 Thunderbolt 4 controller because it combines the best feature from USB, the new Multi-Port Accessory Architecture with the Daisy Chain Accessory Architecture in one product. Also having full bandwidth when a Thunderbolt 4 devices is connected to a USB4 40 host with Thunderbolt 3 compatibility which gives a 360 degrees full circle relationship.
My ideal Thunderbolt 4 dock includes 4x Thunderbolt 4/USB4 USB-C ports (1UP+3DWN), 2 Display (HDMI 2.0 and/or DisplayPort 1.4), 4x USB 3.2 USB-A with quick charge, 1x 10GbE with AVB support.
Moving forward Intel will likely have Thunderbolt integrated to desktop CPUs just as they did with the mobile CPUs like Ice Lake and Tiger Lake. However it will be great to see AMD motherboards with a Maple Ridge Controller to add Thunderbolt 4 and USB4 support. It will be great to see ARM support thunderbolt on something like the Surface Pro X. It will be even better if Microsoft ditched their surface connect and just used Thunderbolt 4 on their next Surface Computers.
I can't wait for Intel to move Thunderbolt to PCIe 4.0 and DisplayPort 2.0; PCIe 3.0 and DisplayPort 1.4 are already bottlenecks.
Deicidium369 - Sunday, December 27, 2020 - link
Rocket Lake will use something similar to this Maple Ridge - not integrated on the CPU. One can assume that Alder Lake will have it integrated inKimGitz - Monday, December 28, 2020 - link
Many 500 series Intel motherboards will feature the Maple Ridge. Outside Apple's M1 Thunderbolt integration, there have been no other 3rd party Thunderbolt controllers announced. Gigabyte have historically led the way for Intel motherboards with Thunderbolt support. AsRock on the other hand took the lead with support for Thunderbolt on their AMD motherboards.Alder Lake will most probably be the 1st generation to feature Thunderbolt integrated to both Desktop and Laptop CPUs. I'm just hoping it will also means upgrading to PCIe 4.0.
lilkwarrior - Saturday, December 26, 2020 - link
Good; hopefully someone makes a Thunderbolt 4 KVM Switch w/ this.KimGitz - Monday, December 28, 2020 - link
With Thunderbolt 4 we get wake up PC feature using a connected mouse or keyboard on a hub/dock, which is a great potential feature for a KVM Switch.WaltC - Tuesday, December 29, 2020 - link
Will this be the Thunderbolt version that fixes all the bugs of the last revision of Thunderbolt?KimGitz - Thursday, December 31, 2020 - link
We can expect every Thunderbolt 4 Host to act the same regardless of platform. We can expect Thunderbolt 4 devices to have better cross platform performance. We can expect the Thunderbolt 4 cable to work regardless of protocol or platform. It really is unfortunate that the first ARM Based Apple M1 Silicon doesn't get Thunderbolt 4 certification mainly due to only supporting one external display and maybe no eGPU support.The is only one thing Thunderbolt 4 doesn't fix is BANDWIDTH.
Thunderbolt 4 is still PCIe 3.0 x4 and 2x DisplayPort 1.4, which is already a bottleneck with PCIe 4.0 SSDs and 8K HDR 60Hz Display Panels.