Intel has introduced a new family of Thunderbolt 3 controllers that bring support of the DisplayPort 1.4 standard to TB3 ports. The new Titan Ridge family of controllers pick up where Intel's previous Alpine Ridge controllers left off by incorporating new DisplayPort functionality, and for the first time, a USB-C fallback mode when used as a sink/peripheral device. This mid-generation update for Thunderbolt 3 will allow the standard and devices using the new Titan Ridge controllers to catch up with current display standards, and work better with the next generation of UHD displays.

Intel’s JHL7x40 family of Thunderbolt 3 controllers supports two main features of the TB3 technology, including PCIe 3.0 with 40 Gbps data transfer rate as well as USB 3.1 Gen 2 with 10 Gbps data transfer rate. The big difference is that Titan Ridge adds support for allowing two DisplayPort 1.4 streams to be encapsulated into the TB3 connection, versus two DisplayPort 1.2 streams in case of the previous-gen TB3 controllers. What isn't changing here is the actual TB3 signaling standard or the cabling, so the total amount bandwidth offered by the previous-gen Alpine Ridge controllers and the new-gen Titan Ridge chips is the same.

The shift in DisplayPort standards is a small but important one for Intel and TB3 device manufacturers. A single DP 1.2 stream is enough for 4K@60Hz displays, but for TB3-enabled monitors beyond that – such as 5K monitors or 4K HDR monitors – vendors would have to resort to multi-tile monitor configurations. Which certainly works for early adopter products, but it's undesirable in the long run due to the higher costs and configuration hassles of a multi-tile monitor. So as these types of monitors become more mainstream and pure DisplayPort monitors shift over to DP 1.4, Thunderbolt 3 has needed to catch up.

Meanwhile, because DP 1.4 has greater bandwidth requirements, it's worth nothing that TB3 displays incorporating Titan Ridge and DP1.4 still cannot exceed 40 Gbps offered by TB3. Formally, one DP 1.4 stream can carry 25.92 gigabits of data per second (32.4 Gbps with overhead) and can support a 5Kp60/8Kp30 display without compression, or a 5Kp120/8Kp60 monitor when the Display Stream Compression 1.2 (DSC) technology is used. However, since in case of the TB3 there is a bandwidth limitation, it will not be possible to plug, say, two 4Kp120 monitors to a single TB3 port on a laptop, despite the fact that Titan Ridge can carry two DisplayPort 1.4 streams. At the same time, something like a single 4Kp144 is now a theoretical possibility (at least for systems with a dGPU). Obviously, since the Thunderbolt 3 technology supports optional USB Power Delivery 3.0 technology, a TB3-enabled laptop can be powered by its TB3 display connection, greatly simplifying setups.

DisplayPort Signaling Standards
Standard Raw Bandwidth
(4 Lanes)
Target Monitor Resolutions
HBR1 (DP 1.0/1.1) 10.8 Gbps 1440p@60Hz
HBR2 (DP 1.2) 21.6 Gbps 4K@60Hz
HBR3 (DP 1.3/1.4) 32.4 Gbps 4K@120Hz &
8K@60Hz (w/DSC)

It's also worth noting that since the DP 1.4 spec is not supported by Intel's iGPUs, Intel-powered notebooks and desktops looking to take advantage of Titan Ridge's DP 1.4 functionality will have to use dGPUs to drive their TB3 controllers. This will somewhat increase the complexity of these designs, since previously most vendors only needed to route the iGPU to the TB3 controllers.

Overall the Titan Ridge family of Thunderbolt 3 controllers consists of three parts: the JHL7540 and the JHL7340 chips for PCs, as well as the JHL7440 chip designed for peripherals. The JHL7340 and the JHL7540 support one and two TB3 ports, respectively, and work exactly like their Apine Ridge predecessors but with the addition of DisplayPort 1.4 functionality.

The JHL7440 however is much more interesting. It too is a dual-port controller, but it is designed specifically for peripherals and is intended to enable compatibility between TB3 peripherals and USB-C hosts. Today’s TB3 docks and displays only work with systems featuring Thunderbolt 3 ports, as there's no way for Alpine Ridge controllers to fall back to being USB-C sinks. By contrast, the JHL7440 controller can fall back for use as a USB-C sink, allowing it to offer "basic compatibility" with USB-C ports. This of course gets into matters such as USB-C's many alt modes – particularly DisplayPort Alt Mode – by at the end of the day the idea is that with Titan Ridge, TB3 peripherals can be connected to a USB-C host and retain most of their functionality. So think TB3 drive arrays that will still operate with less bandwidth, TB3 monitors that may slow down or lose non-display features, etc.

As for why Intel is making this move now, it's worth pointing out that Intel plans to make the Thunderbolt 3 spec available to the industry under a nonexclusive, royalty-free license this year. So it makes a lot of sense for Intel to maximize the compatibility of TB3 peripherals before third-party TB3 controllers emerge.

Intel's Thunderbolt 3 Controllers
  DSL
6240
DSL
6340
DSL
6540
JHL
6340
JHL
6540
JHL
7340
JHL
7440
JHL
7540
Family Alpine Ridge Titan Ridge
Launch Date Q2 2016 Q3 2015 Q2 2016 Q1 2018
TDP 1.2 W 1.7 W 2.2 W 1.7 W 2.2 W ? ? ?
Number of Ports 1 2 1 2 1 2
DisplayPort 2x 1.2 2x 1.4
Package Size 10.7 × 10.7 mm ? ? ?
Price $6.45 $8.00 $8.55 $8.00 $8.55 ? ? ?

Intel’s Titan Ridge controllers will continue to require an external USB Type-C multiplexer and a PD 3.0 controller. Meanwhile, if today’s solutions use the TI TPS65983B chip, there will be other options for the Titan Ridge.

Intel has not published pricing of the new Titan Ridge controllers and intends to do so later in the quarter. As for the availability of devices featuring the new controllers, I suspect we're going to see them sooner than later. Intel’s TB3 software has actually supported the new controllers since mid-2017, hinting that the development of actual hardware by Intel’s partners should be well under way by now.

Related Reading

Source: Intel

Comments Locked

22 Comments

View All Comments

  • DanNeely - Monday, January 8, 2018 - link

    yes, but in USB-C mode it won't be able to do everything all at once.

    Doesn't make much sense for a full up TB3 dock because they run so much more than a basic USB3 one that if you're not using TB3 they're stupidly overpriced. Where it does make sense would be a monitor that's 2.5k/4k+4k passthough+USB3 dock in TB mode being a 2.5/4k monitor with a USB3 (USB2?) dock over USB; or a DAS with an SSD cache that goes from fast to ludicrously fast switching from USB-C to TB3.
  • KimGitz - Thursday, December 13, 2018 - link

    Well I have been in a situation where a client comes in to the studio with the computer but can not use my gear because everything is attached via a Thunderbolt 3 Dock to my computer. Their laptop doesn't have enough USB ports so I can't even unplug stuff from the dock and connect them directly to the clients computer. My expensive dock became useless at that point.
  • Santoval - Monday, January 8, 2018 - link

    I am not sure I understand why the bandwidth was not affected by the switch from DP 1.2 to DP 1.4, while the two streams were retained. Does this mean that the two DP 1.4 streams are nominal rather than full, because they are bandwidth limited by what the four PCIe 3.0 lanes can feed them? And the sole benefit is protocol-specific, due to the support of single stream-5K and HDR?

    By the way, is it me or has DP's 8b/10b encoding been too long in the tooth already? A waste of 6.5 Gbps per stream, or 13 Gbps per pair of streams due to overhead is insane! It's even more insane if you take into account that your PCIe 3.0 x4 links underneath *do* support 128b/130b.
  • repoman27 - Monday, January 8, 2018 - link

    The Titan Ridge Thunderbolt controller has a PCIe 3.0 x4 link to the host for data, and two DP 1.4 sinks. Each of the Thunderbolt ports when operating in Thunderbolt 3 signaling mode provides 2 full-duplex channels at 20.625 Gbit/s with 64b/66b encoding that are bonded, providing 40 Gbit/s to the upper layers. Protocol converters take the PCIe and DP packets and route them over the Thunderbolt links using a series of crossbar switches.

    A 4-lane DisplayPort HBR3 main link bonds 4 simplex channels at 8.1 Gbit/s with 8b/10b encoding for a total of 25.92 Gbit/s, but only in one direction. Also, the link rate is scaled to accomodate the bandwidth required by the display device, and the rest is bit-stuffed. The Thunderbolt protocol converters ditch the stuffing and only put the actual display data packets on the Thunderbolt link. So a Thunderbolt link can carry any number of display streams so long as the aggregate bandwidth required doesn’t exceed 40 Gbit/s, and whatever bandwidth isn’t used for DisplayPort packets can be used for PCIe data.

    The PCIe 3.0 x4 back end is entirely separate from the DisplayPort sink / source connections. It provides 4 full-duplex lanes at a nominal 8 Gbit/s with 128b/130b encoding for 31.5 Gbit/s total. When channel-bonding was introduced with Thunderbolt 2, either the PCIe protocol overhead doubled or the maximum TLP payload size was halved. The net result being that a single Thunderbolt 3 link tops out just shy of 2,750 MB/s for real-world PCIe throughput. That’s limited by the back-end though, not the Thunderbolt link.
  • BillBear - Monday, January 8, 2018 - link

    I will be interested in seeing if target display mode will finally appear for 5K iMacs when these controllers come into use.
  • Poik - Tuesday, January 9, 2018 - link

    Any news on some upcoming AMD boards adopting some TB3 action? Threadripper certainly has enough lanes to do so.
  • Vidmo - Tuesday, January 9, 2018 - link

    It's too bad that Intel won't support Windows server with its Thunderbolt drivers.
  • iwod - Tuesday, January 9, 2018 - link

    While I know TB 3 is awesome tech, i could never understand,

    1. Why are the controller so expensive, $8? There are low end Android Phone SoC selling for $5!
    Which makes the Apple Lightening cable $1.5 chips looks cheap.

    2. Why isn't the USB Multiplexer built in ?

    As a matter of fact why isn't USB 3.1 and TB3, built into same controller?
  • repoman27 - Tuesday, January 9, 2018 - link

    The tray prices for Thunderbolt controllers are actually extraordinarily inexpensive, and I don’t believe they accurately reflect the total cost of incorporating the solution. Show me any other 4-channel I/O transceiver with 20.625 GT/s lane rates that can be had for less than $150. Thunderbolt is bonkers. Lightning cables are just USB 2.0 cables with a proprietary connector on one end. So you’re comparing a chip capable of 1x 480 Mbit/s to one that can do 4x 20.625 Gbit/s.

    Thunderbolt 3 controllers do include the USB 3.1 xHCI on die. The external multiplexer / PMIC is required to support USB PD 2.x and by extension USB Type-C Alternate Modes. This is not a one-size-fits-all solution, requires IP Intel may not care to license, and is generally produced on a different process. So it makes more sense to keep it separate.
  • acana79 - Friday, January 26, 2018 - link

    Why not just add two TB3 controllers to get around the bandwidth issue if we need two monitors for 8K?

Log in

Don't have an account? Sign up now