Details of Intel's Barlow Ridge Thunderbolt 5 Controller Leaks

[TheLostSwede]

Serial leaker @yuuki_ans on X/Twitter has released details on Intel's upcoming Barlow Rridge Thunderbolt 5 controller which will be known as the JHL9580 or JHL9540 depending on the SKU. The good news is that Intel has finally moved to PCIe 4.0 for the bus interface, which was expected due to the increased bandwidth on offer by Thunderbolt 5 over Thunderbolt 3 and 4. Barlow Ridge will use a PCIe 4.0 x4 interface to connect to the host and it appears that the earlier leak that suggested native Thunderbolt support in Arrow Lake-S might be incorrect, as there are diagrams showing Barlow Ridge connected to Arrow Lake-S CPUs.

Besides the faster bus, Thunderbolt 5 brings asymmetrical data transmission support which means that for display applications there will be a 120/40 Gbps mode, whereas for data only applications Thunderbolt 5 will deliver a symmetrical 80 Gbps mode. We should point out that this only appears to apply to the JHL9580 SKU, which also supports 40 Gbps USB4 speeds, whereas the JHL9540 for some reason remains a Thunderbolt 4 controller. That said, both of the Barlow Ridge SKUs get support for 20 Gbps USB 3.2 Gen 2x2, something that was lacking in previous Thunderbolt implementations. There will also be support for DisplayPort 2.1 via DP Alt Mode with full UHBR20 support when used with a DP80 certified cable. Actual data transfers are limited to the 64 Gbps PCIe 4.0 interface to the host system, just like USB4, but this does at least give Thunderbolt 5 extra head room for display data even in symmetrical mode. The Barlow Ridge controllers appear to be connected directly to the Arrow Lake-S CPUs via the PCIe 4.0 x4 interface, much in the same way USB4 host controllers connect to AMD's Ryzen 7000-series CPUS.



View at TechPowerUp Main Site | Source

 

[eldon_magi]

And the max supported cable length will be about 10cm.

 

[Assimilator]

That second slide is so blurred as to be useless.

The good news is that Intel has finally moved to PCIe 4.0 for the bus interface

... not PCIe 5.0? Seriously Intel?

whereas the JHL9540 for some reason remains a Thunderbolt 4 controller

So now USB features are decoupled from the Thunderbolt version, making the TB ecosystem a dumpster fire of confusion along the lines of the USB one. Again, WHY INTEL?

 

[TheLostSwede]

And the max supported cable length will be about 10cm.

1 meter for passive cables.

 

[Onasi]

So now USB features are decoupled from the Thunderbolt version, making the TB ecosystem a dumpster fire of confusion along the lines of the USB one. Again, WHY INTEL?

Remember when the whole goal of standardized connector specs was to make it easy for the user to understand which devices and ports run at which speed and have which features? Good times. We might as well go back to the whole “serial, parallel, game port, printer port, FireWire, etc” mess at this point. I kid… mostly.

 

[TheLostSwede]

That second slide is so blurred as to be useless.

Yeah, agreed, but not my fault.

... not PCIe 5.0? Seriously Intel?

I guess they copied ASMedia...

So now USB features are decoupled from the Thunderbolt version, making the TB ecosystem a dumpster fire of confusion along the lines of the USB one. Again, WHY INTEL?

Well, USB4 is only compatible with Thunderbolt 3, but Thunderbolt 5 is apparently compatible with USB4...
At least Intel finally added support for 20 Gbps USB 3.2 Gen 2x2.

 

[Onasi]

At least Intel finally added support for 20 Gbps USB 3.2 Gen 2x2.

I miss the FullSpeed, HighSpeed and SuperSpeed vernacular. At least it was catchy.
Now… “Ah yes, this port is a USB three point two generation two and two by two port”. Very cool, thank you USB-IF.

 

[eldon_magi]

1 meter for passive cables.

Great! and if you need 2 meter or more, you can get a optical cable handmade in some chinese sweatshop that adheres to no standards, for only $800!

 

[pressing on]

I'm surprised that the details of Intel's Barlow Ridge (not Bridge) Thunderbolt 5 controller are being described as a leak. Intel launched Barlow Ridge/Thunderbolt 5 and supplied full technical details at a developers/press launch in September 2023.

The Intel 14th Gen HX mobile CPUs announced a couple of weeks ago feature Thunderbolt 5 support. At the launch in September there was a working prototype Gigabyte HX laptop using Barlow Ridge.

Some of the "leaked" documentation looks like it could have been adapted from the technical data supplied to the laptop companies for the HX chips. For example "Arrow Lake S DDR5 SODIMM" - SODIMMs are used by laptops not desktops?

 

[TheLostSwede]

Great! and if you need 2 meter or more, you can get a optical cable handmade in some chinese sweatshop that adheres to no standards, for only $800!

No need for optical, you just need an active copper cable. They're about $40 for 2 meters. Although there are no Thunderbolt 5 cables yet, so I'm going by USB4 pricing.

https://www.amazon.com/Charging-Compatible-Thunderbolt-Transfer-Charger/dp/B0BNL1WGWJ/

I'm surprised that the details of Intel's Barlow Ridge (not Bridge) Thunderbolt 5 controller are being described as a leak. Intel launched Barlow Ridge/Thunderbolt 5 and supplied full technical details at a developers/press launch in September 2023.

Thanks, I was clearly not awake this morning when I wrote it, so my brain wanted it to be Bridge... Also, the redrivers/retimers are Hayden Bridge, which caused further brain confusion on my behalf.
No public details are available though, hence why it's a leak, since these are clearly from confidential documents.

The Intel 14th Gen HX mobile CPUs announced a couple of weeks ago feature Thunderbolt 5 support. At the launch in September there was a working prototype Gigabyte HX laptop using Barlow Ridge.

I was referring to this earlier leak, but it still looks like native Thunderbolt support, but only up to 40 Gbps. Maybe we'll get both? I guess even this could be Thunderbolt 5 based on what's mentioned in the Barlow Ridge specs.

Detailed Intel Arrow Lake-S Platform Specifications Leaked, Confirms Native Thunderbolt 4 Support

Courtesy of X/Twitter user @yuuki_ans, we now have what should be very detailed information on Intel's next generation consumer desktop platform, assuming the leaked information is real. The leaker not only provided confirmation on the CPU specs of the Arrow Lake-S desktop CPUs, which will...

www.techpowerup.com

Some of the "leaked" documentation looks like it could have been adapted from the technical data supplied to the laptop companies for the HX chips. For example "Arrow Lake S DDR5 SODIMM" - SODIMMs are used by laptops not desktops?

True, though some industrial/embedded boards use them too.

 

[Tek-Check]

The Intel 14th Gen HX mobile CPUs announced a couple of weeks ago feature Thunderbolt 5 support. At the launch in September there was a working prototype Gigabyte HX laptop using Barlow Ridge.

There was, but they did not show one single parametre of this connection.
I am looking now on Gigabyte's website to check specs for 17X laptops with HX CPUs. None of them has TB5 port. It's nonsense.
The problem is that Nvidia GPUs do not support DP 2.1 but only older DP 1.4.
Where is Barlow Ridge supposed to take 80 Gbps video signal from? From iGPU on $3,500 laptop?

 

[pressing on]

There was, but they did not show one single parametre of this connection.
I am looking now on Gigabyte's website to check specs for 17X laptops with HX CPUs. None of them has TB5 port. It's nonsense.
The problem is that Nvidia GPUs do not support DP 2.1 but only older DP 1.4.
Where is Barlow Ridge supposed to take 80 Gbps video signal from? From iGPU on $3,500 laptop?

You're right that the current Thunderbolt 5 support that Intel has announced simply means that there is provision for an optional Barlow Ridge controller to be installed - no Barlow Ridge, no TB5. The only TB5 laptop that comes to mind immediately is the recently announced Razor Blade 18 that Razor say includes "...1 x Thunderbolt™ 5 (USB4®V2) Type-C with Power Delivery and Display Port 1.4 with Nvidia GPU output."

The Razor Blade 18 has a 14th gen Raptor Lake processor, it is likely that around this time next year there will be an Arrow Lake replacement for the current HX mobile chips. It may be that TB5 will be standard rather than an option for the AR mobile processors, whether this will also extend to desktop Arrow Lake is not clear at this stage.

The source of the graphic below is Intel from the launch for the Core 14th Gen HX mobile processors in January

 

[Tek-Check]

We should point out that this only appears to apply to the JHL9580 SKU, which also supports 40 Gbps USB4 speeds, whereas the JHL9540 for some reason remains a Thunderbolt 4 controller. That said, both of the Barlow Ridge SKUs get support for 20 Gbps USB 3.2 Gen 2x2, something that was lacking in previous Thunderbolt implementations. There will also be support for DisplayPort 2.1 via DP Alt Mode with full UHBR20 support when used with a DP80 certified cable.

Leaked information and currently implemented TB5 solutions are somewhat confusing, and I just do not understand why this is the case. Why has Intel not simply presented all Barlow Ridge SKUs and transparently outlined specifictation of each SKU? It's mind-boggling...

It seems to me that we might have three main SKUs, like in previous generations:
- 'JHL9380' - not yet seen, single port, but it might emerge at some point...
- JHL9580 - dual port TB5 controller with symetrical Tx/Rx 80 Gbps, plus dynamic Tx/Rx allocation 120/40 Gbps
- JHL9480 - quad port chip in upcoming docks, 1-in-3-out, symetrical Tx/Rx 80 Gbps and dynamic Tx/Rx 120/40 Gbps
https://zhuanlan-zhihu-com.translat...uto&_x_tr_tl=en&_x_tr_hl=en-US&_x_tr_pto=wapp

*JHL9540 - this one is so weird; the leak you mentioned suggests refreshed TB4 controller that could be also attached to retimer Hayden Bridge to allow UHBR10 bandwidth at 40 Gbps from CPU or GPU, mainly on laptops, but also from desktop Arrow Lake CPU, as shown above. Puzzling chip, indeed.

Below are all TB5 docks annouced at CES, as you probably know. Those should feature linked quad port controller JHL9480, unless there is an unlikely hybrid solution with two JHL9580 chips on PCB.
TB5 dock - HyperDrive
TB5 dock - Good Way DBF7050
TB5 dock - j5create JTD5172
TB5 dock - Kensington SD5000T5
TB5 dock - IOGear Quantum Pro GTD58K3TAA
TB5 dock - OWC

The only TB5 laptop that comes to mind immediately is the recently announced Razor Blade 18 that Razor say includes "...1 x Thunderbolt™ 5 (USB4®V2) Type-C with Power Delivery and Display Port 1.4 with Nvidia GPU output."

I looked into this too. This spec from Razer is so confusing. How can there be Thunderbolt 5 port with DP 1.4 signal? It just does not make sense, unless Intel has watered down the spec of TB5. The only way to currently feed Barlow Ridge controller with DP 2.1 video from dGPU is from AMD 7000 cards. No other discrete GPU product supports DP 2.1

The Razor Blade 18 has a 14th gen Raptor Lake processor, it is likely that around this time next year there will be an Arrow Lake replacement for the current HX mobile chips. It may be that TB5 will be standard rather than an option for the AR mobile processors, whether this will also extend to desktop Arrow Lake is not clear at this stage.

ARL will have integrated TB4. For TB5 integreation into iGPU, we will need to wait until either Panther Lake or later.

The source of the graphic below is Intel from the launch for the Core 14th Gen HX mobile processors in January

I know this one, but the graphic is just marketing... it says nothing about real solutions. And current solutions look so confusing and convoluted.

 

[kimgitz]

Leaked information and currently implemented TB5 solutions are somewhat confusing, and I just do not understand why this is the case. Why has Intel not simply presented all Barlow Ridge SKUs and transparently outlined specifictation of each SKU? It's mind-boggling...

It seems to me that we might have three main SKUs, like in previous generations:
- 'JHL9380' - not yet seen, single port, but it might emerge at some point...
- JHL9580 - dual port TB5 controller with symetrical Tx/Rx 80 Gbps, plus dynamic Tx/Rx allocation 120/40 Gbps
- JHL9480 - quad port chip in upcoming docks, 1-in-3-out, symetrical Tx/Rx 80 Gbps and dynamic Tx/Rx 120/40 Gbps
https://zhuanlan-zhihu-com.translat...uto&_x_tr_tl=en&_x_tr_hl=en-US&_x_tr_pto=wapp

*JHL9540 - this one is so weird; the leak you mentioned suggests refreshed TB4 controller that could be also attached to retimer Hayden Bridge to allow UHBR10 bandwidth at 40 Gbps from CPU or GPU, mainly on laptops, but also from desktop Arrow Lake CPU, as shown above. Puzzling chip, indeed.

Below are all TB5 docks annouced at CES, as you probably know. Those should feature linked quad port controller JHL9480, unless there is an unlikely hybrid solution with two JHL9580 chips on PCB.
TB5 dock - HyperDrive
TB5 dock - Good Way DBF7050
TB5 dock - j5create JTD5172
TB5 dock - Kensington SD5000T5
TB5 dock - IOGear Quantum Pro GTD58K3TAA
TB5 dock - OWC


I looked into this too. This spec from Razer is so confusing. How can there be Thunderbolt 5 port with DP 1.4 signal? It just does not make sense, unless Intel has watered down the spec of TB5. The only way to currently feed Barlow Ridge controller with DP 2.1 video from dGPU is from AMD 7000 cards. No other discrete GPU product supports DP 2.1

ARL will have integrated TB4. For TB5 integreation into iGPU, we will need to wait until either Panther Lake or later.

I know this one, but the graphic is just marketing... it says nothing about real solutions. And current solutions look so confusing and convoluted.

From the leak the Barlow Ridge controller is rated at 3.2-4W. What is the rating for Maple Ridge and Goshen Ridge?

I have heard rumours that TB 5 controllers is difficult to cool. I don't think it is true. I wonder if you have any information on this.
I think the biggest problem is the requirement for DP 2.1 for Thunderbolt 5.
Lunar Lake now supports eDP1.5, DP 2.1 and HDM1 2.1. The Intel Arc Xe2 Battlemage discrete GPU will also support DP 2.1.
I'm really disappointed by the launch of Thunderbolt 5 it has been poorly executed.

 

[Tek-Check]

From the leak the Barlow Ridge controller is rated at 3.2-4W. What is the rating for Maple Ridge and Goshen Ridge?

Intel did not publish power ratings.

I have heard rumours that TB 5 controllers is difficult to cool. I don't think it is true. I wonder if you have any information on this.

Nothing at the moment. Probably need to look more closely into it when I get some time.

I think the biggest problem is the requirement for DP 2.1 for Thunderbolt 5.

Apparently, it's not, because Razer released laptop with "TB5" featuring DP 1.4a signal only.
DP80 sources include Meteor Lake, Lunar Lake and AMD Pro 7000 cards. It can be done properly.

I'm really disappointed by the launch of Thunderbolt 5 it has been poorly executed.

Me too.