Earlier this month the USB Implementers Forum announced the new USB Power Delivery 2.0 specification. Long awaited, the Power Deliver 2.0 specification defined new standards for power delivery to allow Type-C USB ports to supply devices with much greater amounts of power than the previous standard allowed, now up to 5A at 5V, 12V, and 20V, for a maximum power delivery of 100W. However also buried in that specification was an interesting, if cryptic announcement regarding USB Alternate Modes, which would allow for different (non-USB) signals to be carried over USB Type-C connector. At the time the specification simply theorized just what protocols could be carried over Type-C as an alternate mode, but today we finally know what the first alternate mode will be: DisplayPort.

Today the VESA is announcing that they are publishing the “DisplayPort Alternate Mode on USB Type-C Connector Standard.” Working in conjunction with the USB-IF, the DP Alt Mode standard will allow standard USB Type-C connectors and cables to carry native DisplayPort signals. This is designed to open up a number of possibilities for connecting monitors, computers, docking stations, and other devices with DisplayPort video while also leveraging USB’s other data and power capabilities. With USB 3.1 and Type-C the USB-IF was looking to create a single cable that could carry everything, and now that DisplayPort can be muxed over Type-C, USB is one step closer to that with the ability to carry native video.

The Tech & The Spec

From a technical level the DP Alt Mode specification is actually rather simple. USB Type-C – which immediately implies using/supporting USB 3.1 signaling – uses 4 lanes (pairs) of differential signaling for USB Superspeed data, which are split up in a 2-up/2-down configuration for full duplex communication. Through the Alt Mode specification, DP Alt Mode will then in turn be allowed to take over some of these lanes – one, two, or all four – and run DisplayPort signaling over them in place of USB Superspeed signaling. By doing so a Type-C cable is then able to carry native DisplayPort video alongside its other signals, and from a hardware standpoint this is little different than a native DisplayPort connector/cable pair.

From a hardware perspective this will be a simple mux. USB alternate modes do not encapsulate other protocols (ala Thunderbolt) but instead allocate lanes to those other signals as necessary, with muxes at either end handling the switching to determine what signals are on what lanes and where they need to come from or go. Internally USB handles this matter via the CC sense pins, which are responsible for determining cable orientation. Alongside determining orientation, these pins will also transmit a Standard IDentification (SID), which will be how devices negotiate which signals are supported and which signals to use. After negotiation, the devices at either end can then configure themselves to the appropriate number of lanes and pin orientation.

Along with utilizing USB lanes for DP lanes, the DP Alt Mode standard also includes provisions for reconfiguring the Type-C secondary bus (SBU) to carry the DisplayPort AUX channel. This half-duplex channel is normally used by DisplayPort devices to carry additional non-video data such as audio, EDID, HDCP, touchscreen data, MST topology data, and more. Somewhat perversely in this case, the AUX channel has even been used to carry USB data, which dutifully enough would still be supported here for backwards compatibility purposes.

Since the main DisplayPort lanes and AUX channel can be carried over Type-C, when utilized in this fashion Type-C is very close to becoming a superset of DisplayPort. In a full (4 lane) DisplayPort configuration, along with all of the regular DisplayPort features a Type-C cable also carries the standard USB 2.0 interface and USB power, which always coexist alongside alt mode. So even in these configurations Type-C allows dedicated high power and USB 2.0 functionality, something the DisplayPort physical layer itself is not capable of. And of course when using a less-than-full configuration, 2-3 of those lanes on the Type-C cable then can be left to running USB Superspeed signaling, allowing USB 3.1 data to be carried alongside the narrower DisplayPort signal.

Meanwhile since DP Alt Mode means that Type-C carries native DisplayPort signaling, this enables several different interoperability options with other Type-C devices and legacy DisplayPort devices. On the hardware side Type-C ports can be used for the sink (displays) as well as the source (computers), so one could have a display connected to a source entirely over Type-C. Otherwise simple Type-C to DisplayPort cables can be constructed which from the perspective of a DisplayPort sink would be identical to a real DisplayPort cable, with the cable wired to expose just the DisplayPort signals to the sink. Or since these cables will be bidirectional, a legacy DisplayPort source could be connected to a Type-C sink just as well.

This also means that since DP Alt Mode is such a complete implementation of DisplayPort, that DisplayPort conversion devices will work as well. DisplayPort to VGA, DVI, and even HDMI 2.0 adapters will all work at the end of Type-C connection, and the VESA will be strongly encouraging cable makers to develop Type-C to HDMI 2.0 cables (and only HDMI 2.0, no 1.4) to make Type-C ports usable with HDMI devices. In fact the only major DisplayPort feature that won’t work over a Type-C connector is Dual-Mode DisplayPort (aka DP++), which is responsible for enabling passive DisplayPort adapters. So while adapters work over Type-C, all of them will need to be active adapters.

From a cabling standpoint DP Alt Mode will have similar allowances and limitations as USB over Type-C since it inherits the physical layer. DisplayPort 1.3’s HBR3 mode will be supported, but like USB’s Superspeed+ (10Gbps) mode this is officially only specified to work on cables up to 1M in length. Meanwhile at up to 2M in length DisplayPort 1.2’s HBR2 mode can be used. Meanwhile DP Alt Mode is currently only defined to work on passive USB cables, with the VESA seemingly picking their words carefully on the use of “currently.”

The Ecosystem & The Future

Because of the flexibility offered through the DP Alt Mode, the VESA and USB-IF have a wide range of options and ideas for how to make use of this functionality, with these ideas ultimately converging on a USB/DisplayPort ecosystem. With the ability to carry video data over USB, this allows for devices that make use of both in a fashion similar to Thunderbolt or DockPort, but with the greater advantage of the closer cooperation of the USB-IF and the superior Type-C physical layer.

At its most basic level, DP Alt Mode means that device manufacturers would no longer need to put dedicated display ports (whether DisplayPort, VGA, or HDMI) on their devices, and could instead fill out their devices entirely with USB ports for all digital I/O. This would be a massive boon to Ultrabooks and tablets, where the former only has a limited amount of space for ports and the latter frequently only has one port at all. To that end there will even be a forthcoming identification mark (similar to DP++) that will be used to identify Type-C ports that are DP Alt Mode capable, to help consumers identify which ports they can plug their displays into. The MUX concept is rather simple for hardware but I do get the impression that devices with multiple Type-C ports will only enable it on a fraction of their ports, hence the need for a logo for consumers to identify these ports. But we’ll have to see what shipping devices are like.

More broadly, this could be used to enable single-cable connectivity for laptops and tablets, with a single Type-C cable providing power to the laptop/tablet while also carrying input, audio, video, additional USB data, and more. This would be very similar to the Thunderbolt Display concept, except Type-C would be able to be a true single cable solution since it can carry the high-wattage power that Thunderbolt can’t. And since Type-C can carry DisplayPort 1.3 HBR3, this means that even when driving a 4K@60Hz display there will still be 2 lanes of USB Superspeed+ available for any devices attached to the display. More likely though we’ll see this concept first rolled out in dock form, with a single dock device connecting to an external monitor and otherwise serving as the power/data hub for the entire setup.

Speaking of which, this does mean that USB via DP Alt Mode will more directly be competing with other standards such as Thunderbolt at DockPort. Thunderbolt development will of course be an ongoing project for Intel, however for DockPort this is basically the end of the road. The standard, originally developed by AMD and TI before being adopted by the VESA, will continue on as-is and will continue to be supported over the DisplayPort physical layer as before. However it’s clear from today’s announcement that DisplayPort over USB has beaten USB over DisplayPort as the preferred multi-signal cabling solution, leaving DockPort with a limited duration on the market.

It’s interesting to note though that part of the reason DP Alt Mode is happening – and why it’s going to surpass DockPort – is because of the Type-C physical layer. In designing the Type-C connector and cabling, the USB-IF has specific intentions of having the Type-C connector live for a decade or more, just like USB Type-A/B before it. That means they’ve done quite a bit of work to future-proof the connector, including plenty of pins with an eye on supporting speeds greater than 10Gbps in the future.

For that reason the possibility is on the table of ditching the DisplayPort physical layer entirely and relying solely on Type-C. Now to be clear this is just an option the technology enables, but for a number of reasons it would be an attractive option for the VESA. As it stands the DisplayPort physical layer tops out at 8.1Gbps per lane for HBR3, meanwhile Superspeed+ over Type-C tops out at 10Gbps per lane with the design goal of further bandwidth increases. As the complexity and development costs of higher external buses goes up, one could very well see the day where DisplayPort was merely the protocol and signaling standard for monitors while Type-C was the physical layer, especially since DisplayPort and USB Superspeed are so very similar in the first place due to both using 4 lanes of differential signaling. But this is a more distant possibility; for now the DP Alt Mode ecosystem needs to take off for the kinds of mobile devices it’s designed for, and only then would anyone be thinking about replacing the DisplayPort physical layer entirely.

Wrapping things up, the VESA tells us that they are going to hit the ground running on DP Alt Mode and are seeing quite a bit of excitement from manufacturers. The VESA is expecting the first DP Alt Mode capable devices to appear in 2015, which is the same year Type-C ports begin appearing on devices as well. So if everything goes according to schedule, we should see the first DP Alt Mode devices in just over a year.

The all-in-one cable concept has been a long time coming, and after DockPort and Thunderbolt stumbling the market does look ripe for DP Alt Mode. So long as the execution is there, the manufacturers are willing to use it, and device compatibility lives up to the promises. Getting video over USB is the ultimate Trojan horse – unlike mDP, USB is already everywhere and will continue to be – so this may very well be the X factor needed to see widespread adoption where other standards have struggled.

POST A COMMENT

117 Comments

View All Comments

  • Ryan Smith - Monday, September 22, 2014 - link

    That's 8.4Gbps per pane. Just as on DP, Type-C allows you 1-4 lanes.

    So Type-C can support 5K@60Hz just as well as DP 1.3 can.
    Reply
  • B3an - Monday, September 22, 2014 - link

    That is awesome. This is the perfect connector, finally! I was looking forward to DP 1.3 because HDMI 2.0 is a disappointment, and i want a 4K or 5K monitor. But now DP seems kind of pointless, i'd rather they just use Type C on everything. Reply
  • SirKnobsworth - Monday, September 22, 2014 - link

    That's per link. DisplayPort has four parallel links, though it can function with fewer. Reply
  • The Von Matrices - Monday, September 22, 2014 - link

    It's increasingly hard to justify the "serial" part of USB. Now that the type C cable has 5 wire pairs (up from three in USB 3.0 and one in USB 2.0), when does it become the Universal Parallel Bus? Reply
  • SirKnobsworth - Tuesday, September 23, 2014 - link

    DisplayPort and PCIe can bond multiple channels together into a single link, but I think they're still considered serial because individual bytes are transmitted over a single link at a time rather than each bit being sent over a different pin as in the old ATA, SCSI, and Parallel Ports. Reply
  • repoman27 - Tuesday, September 23, 2014 - link

    "USB Type-C – which immediately implies using/supporting USB 3.1 signaling – uses 4 lanes (pairs) of differential signaling for USB Superspeed data, which are split up in a 2-up/2-down configuration for full duplex communication."

    Ryan, please, because this is Anandtech and not one of those other sites, get this right.

    The USB Type-C Cable and Connector Specification does not imply USB 3.1 at all. In fact, it explicitly defines USB 2.0 Type-C receptacles, plugs and cables. Type-C is a replacement for Type-A and all of the many B variants. Although it might be total overkill to use Type-C with its 24-contact connectors and up to 18-conductor cables for USB 2.0 only applications, I think we'll see it anyway simply due to the size and reversibility factors.

    Only 2 differential signaling pairs are ever used for SuperSpeed / SuperSpeedPlus links. Signal paths for 4 lanes (2 Tx/Rx pairs) are provided in Type-C to facilitate mapping the USB signals independent of plug orientation in the receptacle.

    "...2-3 of those lanes on the Type-C cable then can be left to running USB Superspeed signaling..."

    Once again, only 2 lanes (1 TX/RX pair, or a single full-duplex channel) can be used for SuperSpeed signaling.

    Also, USB 3.1 introduces the Enhanced SuperSpeed Bus Architecture which includes both the Gen 1 physical layer and SuperSpeed protocol of USB 3.0 (5 Gbit/s with 8b/10b encoding) and the Gen 2 physical layer and SuperSpeedPlus protocol new to USB 3.1 (10 Gbit/s with 128b/132b encoding). There seems to be a ton of confusion about this which is undoubtedly attributable to the ridiculous marketing names bestowed thus far on the various USB protocols.
    Reply
  • zodiacfml - Tuesday, September 23, 2014 - link

    wow, very impressive. It even includes 4k@60Hz which I really like.
    i thought lighpeak (thunderbolt) will be the be the "it" cable....but just the nuisance of charging our devices makes this cable a must.
    Reply
  • gkujyhtsre - Wednesday, September 24, 2014 - link

    Network? Reply
  • Portme - Wednesday, September 24, 2014 - link

    Displayport Alt Mode?! Did they not notice the acronym? Now I'm going to have to plug my DAM cable into my DAM computer port in order to get a DAM connection to my monitor. DAM it!
    Ok, if we don't misinterpret the spelling, then that leaves the question, does DAM do to my signals what a dam does to a river?
    They picked the worst name since fracking. It's worse than WinCE. They need to pick a different name, pronto. Or at least start saying DPAM a lot, to inoculate the public, to prevent hooligans like me from infecting everybody with the DAM meme.

    More seriously:
    1. The newest SATA is SATA Express, which is simply NVMe-or-AHCI over PCIe (plus a couple of legacy SATA links bundled into the cable), with its own dedicated connector type.
    2. USB with DAM is USB-3 and DP multiplexed onto the same physical link (plus a USB-2 link and power lines bundled into the cable), with a USB-C connector.
    3. Thunderbolt is essentially encapsulated PCIe and DP multiplexed onto the same physical link, with a DP connector.
    4. USB with PCIe altmode will be USB-3 and PCIe multiplexed onto the same physical link, with a USB-C connector.

    All those high-speed physical links are similar in speed per signaling pair, within a factor of 2, depending on the currently-latest version of the spec. #4 will undoubtedly enable use of all four high-speed pairs in USB-C, just as DAM does. #4 will have no more overhead or latency than Thunderbolt does.

    Notice what #1 and #4 together imply: SATA Express will become an altmode for USB-C with no extra effort. Thus when #4 is available, the dedicated connector type for #1 will be unnecessary, and eSATA (external SATA) will be unnecessary. UAS will also be unnecessary. USB-C will be the highest-speed, lowest-latency connection available for disks and SSDs, both internal and external, except for internal SSD cards that plug directly into motherboard PCIe slots.

    Using USB-C for data and power even for internal disks will be very nice for people who put together desktops and servers. Besides providing a physically small connector and eliminating the need for different internal vs. external cable types, this also lets drives be used internally or externally without any adapter needed, which reduces the cost premium for external drives, since the premium is then just for the physically protective housing.

    In a future version of the spec, the USB-3 protocol could itself be made into an optional altmode; after all, if USB-C is used for internal disks, there's no reason to waste money supporting USB-3 for the connector on the motherboard, since only the PCIe mode will be used.

    With USB-C able to carry 40Gbps aggregate via 4 signaling pairs, just as 40GBase-T ethernet does, the latter could be an altmode for the former. Besides providing a small and nicer connector than traditional 8P8C, USB-C also provides much more efficient power transmission than power over 24 AWG as with PoE.
    Reply
  • SirKnobsworth - Thursday, September 25, 2014 - link

    Minor note - ThunderBolt is more than just multiplexing - I believe it aggregates packets from multiple sources and resends them over a higher bandwidth link - TB has 10 Gbit/s links whereas the PCIe 2.0 and DP 1.2 links it feeds from are only ~5 Gbit/s each. TB3 will double that to 20 Gbit/s links - with active cables, of course. Reply

Log in

Don't have an account? Sign up now