Report: Next Thunderbolt chip to double bandwidth, add device charging

Intel’s Thunderbolt interconnect is apparently scheduled for another upgrade. According to an official-looking slide published by VR-Zone’s Chinese alter-ego, a new controller dubbed Alpine Ridge is in the works. This chip promises 40Gbps of bandwidth—double what’s available in the current generation of Falcon Ridge hardware.

The faster interface will enable Alpine Ridge to power dual 4K displays over a single cable, the slide says. PCI Express devices should perform faster, as well. The interconnect’s PCIe link has reportedly been upgraded from Gen2 to Gen3, which doubles the theoretical bandwidth available to individual peripherals. Impressively, these gains haven’t increased the chip’s power consumption. The leaked slide claims a 50% reduction in power versus existing Thunderbolt tech.

Speaking of power, the slide mentions "system charging" support for devices up to 100W. That power rating matches the latest USB Power Deliver spec, and I hope there’s cross-compatibility between the two schemes.

The bandwidth upgrade and charging support require a new connector, according to the slide. The new plug will reportedly be just 3 mm thick, making it better-suited to slim mobile devices. Adapters will apparently be available to ensure backward compatibility with older gear.

There’s no word on when this rumored Thunderbolt upgrade is due to arrive. However, the slide shows Alpine Ridge connected to Skylake, the architectural refresh scheduled after Broadwell. With the first Broadwell CPUs expected late this year and no ETA for Skylake, it could be a while before the next Thunderbolt controller is out in the wild. And that’s just fine, because PC users haven’t really embraced the current Thunderbolt implementation. Perhaps this next iteration will come with more compelling devices, such as external graphics modules for ultraportable notebooks. Fingers crossed.

Comments closed
    • hasseb64
    • 7 years ago

    New standards – new Connections?
    AGAIN!!
    Nuts!

    • Sahrin
    • 7 years ago

    Dear Intel,

    The problem with Thunderbolt is the cost, not the technology.

    Love,

    Everyone but Apple

      • Krogoth
      • 7 years ago

      Thunderbolt was always going to be more expensive than USB. It involves a lot more shielding, tracing and special controller chips to make that level of bandwidth possible over conductive metals.

      10Gbps and 40Gbps Ethernet isn’t that much better unless you are going for short cable runs.

      Simply put, we are running out of room with conductive metals. It is time go to optical (which is what Light Peak was suppose to be).

        • Sahrin
        • 7 years ago

        [quote<]Thunderbolt was always going to be more expensive than USB. It involves a lot more shielding, tracing and special controller chips to make that level of bandwidth possible over conductive metals. [/quote<] Nope. It is more expensive because it uses a dedicated controller. USB makes 10 Gbps without the cost. It's not rocket science, you just have to do it. [quote<]10Gbps and 40Gbps Ethernet isn't that much better unless you are going for short cable runs.[/quote<] 1. 40 Gbps and 10 Gbps belong to two different standards (40 Gbps is part of 100 GBASE). 2. 10GBASE is *not* that much more expensive than 1000Base-TX. [quote<]Simply put, we are running out of room with conductive metals.[/quote<] Nope. There is plenty of bandwidth in the cables, we just need to implement techniques to obtain it. There is no underlying physical limit like you are implying.

    • Chrispy_
    • 7 years ago

    The only headline that would be of interest to the vast majority of the world would be:

    [quote<]Report: Next Thunderbolt chip to reduce cable costs by an order of magnitude[/quote<] Thunderbolt adoption rates are dire, because every time anyone buys a thunderbolt device and needs another cable, they feel scalped and swear to buy a USB device next time.

    • UnfriendlyFire
    • 7 years ago

    The one major application I can see TB shining in is an external GPU to go with an ultrabook. Low weight and long battery on the go, desktop-like gaming when stationary and plugged in.

    The only issue is, the cheapest Thunderbolt to PCI-E external GPU case cost way over $200 the last time I checked.

    Can anyone explain to me how a TB to PCI-E connector, vented aluminum box, and a mini-PSU cost as much as a mid-range GPU itself?

      • smilingcrow
      • 7 years ago

      1. Low volume = high pricing.
      2. TB is primarily used by Apple so users prefer to over pay. Joke.

    • RoninGyrbill
    • 7 years ago

    Could you power your monitor through this? Currently I have 4 ugly thick cables sticking out of the back of my 1440 monitor, this could theoretically reduce it to 1. usb cable, power cable, displayport cable (and hdmi for the ps3 hooked up). I figure most monitors are under 100w even at full brightness.

    • HisDivineOrder
    • 7 years ago

    Thunderbolt is the next generation of Firewire in every way. It really is. It’s like Intel asked themselves, “How can we make a new Firewire to go with the new USB?” all those years ago.

    Thunderbolt has its niche. USB3 owns the rest. Thunderbolt remains too expensive to use in anything except the most expensive of uses anyway, so it’ll remain strongly there. Elsewhere, USB3 will the cheap, good enough experience for the rest of use cases that don’t need or even benefit from the extra that Thunderbolt’s expense offers.

    So Thunderbolt at this point gains nothing from being cheaper. Being better, though, might net it some wins in the niche it’s carved out for itself.

    • crabjokeman
    • 7 years ago

    I’m not paying for Thunderbolt’s garbage licensing fees any more than I paid for pkzip. Maybe intel needs to take a hard look at what makes a successful standard…

    • albundy
    • 7 years ago

    intel really does have alot of money to blow. TB has been a failure since its first incarnation and intel still thinks it can convince consumers that it is worthwhile.

      • Farting Bob
      • 7 years ago

      Apple likes it at least. Intel doesn’t need it to become the new USB (and knows it never will because of it’s active cable requirements, it will never be as cheap), it can still make a tidy profit from niche usage.

      • smilingcrow
      • 7 years ago

      It’s shipped in millions of Apple computers so hardly a failure. PCs usually have more expansion options than Apple’s closed boxes so there’s less of a need for it on PCs so it was always likely to remain niche.

    • zenlessyank
    • 7 years ago

    Might as well tell us its 16X as fast as the old port, since it is just going to sit there unused.
    We hardly have ANY TB devices as it is, and now we get a new one.

    It’s hard enough finding true USB 3 devices.

    Slap on a new updated RS-232 while you are adding useless ports!!!

    coal subscriber.

    • Ninjitsu
    • 7 years ago

    If the HWiNFO v4.37 changelog is anything to go by…
    [quote<]-Enhanced Broadwell and Skylake support.[/quote<] [url<]http://www.hwinfo.com/news.php[/url<] ...Skylake should be here next year too. Probably get a better idea at Computex.

    • UnfriendlyFire
    • 7 years ago

    Hey Intel, how about you make the cables and the tracing implementation cheaper?

    There’s a reason why Firewire lost to USB.

      • the
      • 7 years ago

      Cabling isn’t going to get any cheaper due to the need for them to be active. If the faster spec requires new cabling, I’m starting to think that it would have been wiser to just go with optical as Thunderbolt was originally conceived to do (aka LightPeak).

        • Deanjo
        • 7 years ago

        Active cabling doesn’t need to be expensive. Redmere cables for example are relatively inexpensive.

          • NeelyCam
          • 7 years ago

          HDMI is much slower than Thunderbolt.

            • Deanjo
            • 7 years ago

            Monoprices Redmere cables can already handle 18Gbps (possibly more). So while it is slower than the proposed 40 Gbps, they are already fast enough to match current TB connections and still remain inexpensive.

            • NeelyCam
            • 7 years ago

            Um, that’s 3x6Gbps… It’s not the same thing as Thunderbolt.

            • Deanjo
            • 7 years ago

            Thunderbolt utilizes multiple channels as well.

            • NeelyCam
            • 7 years ago

            Are you intentionally trying to ignore the point? TB channels are operating at much higher data rates than HDMI channels. You can’t expect 20Gbps repeaters to be cheap

            • Deanjo
            • 7 years ago

            Ummm, Alpine Ridge is using 4 channels x 10Gbps.

            • NeelyCam
            • 7 years ago

            It’s predecessor, Falcon Ridge, is already at 2x20Gbps, at least according to [url=http://www.anandtech.com/show/7026/intel-announces-thunderbolt-2-at-computex-20gbps-bidirectional-bandwidth-per-channel<]AnandTech[/url<]. Why would the next-gen go back to 10Gbps/channel? Do you have a link that shows this? Because I can't find any info on Alpine Ridge..

            • mesyn191
            • 7 years ago

            CAT7a/class F gets 40Gbps over regular copper up to 50m. It has a fair amount of shielding so it costs more than CAT6, 5, 5e, etc. but its still pretty cheap in comparison to any active cable.

            • NeelyCam
            • 7 years ago

            Are you talking about this wiki comment?

            [quote<]"Simulation results have shown that 40 Gigabit Ethernet may be possible at 50 meters (164 ft) and 100 Gigabit Ethernet at 15 meters (49 ft).[citation needed] "[/quote<] Simulations != real life, and where's that citation?

            • mesyn191
            • 7 years ago

            Here it is:

            [url<]https://etda.libraries.psu.edu/paper/10174/[/url<] Remember, they're testing network cabling and not a new rocket engine or doing ground breaking physics work. The big issue the paper brings up isn't the cabling, its the controllers necessary to handle those data rates. edit: if you still want a real world example of 40Gbps over copper this is a test of CAT8 doing it: [url<]http://www.ieee802.org/3/NGBASET/public/jan13/Larsen_01a_0113_NGBT.pdf[/url<]

            • NeelyCam
            • 7 years ago

            Yes; that’s pretty interesting work, but rather extreme from coding perspective (=way more expensive than those simple Redmere repeaters). And, they are using 4-way MIMO scheme on the four-channel ethernet cable, so it’s more like 10Gbps per channel – not 40gbps.

            In ISSCC2014, Intel paper discussed a 128Gbps link over a single cable (4 channels, 32Gbps per channel), so yes – huge bandwidth can be driven through an electrical cable. But that stuff is probably quite a bit more expensive than 3x6Gbps Redmere cables, which was my whole point: driving the data rate up per channel isn’t easy or cheap.

            • mesyn191
            • 7 years ago

            The cost will be higher than existing 1Gbps stuff certainly but can be way cheaper this Thunderbolt crap Intel keeps pushing.

            Which is kinda my whole point, though I didn’t state it explicitly, that there are cheaper and more practical ways to get higher bandwidth than Thunderbolt.

            Arguing over per channel bandwidth is kind’ve pointless since the hardware controllers will ‘hide’ that and all the end user is going to be dealing with and see is a single cable a la CAT5/6/etc pushing 40Gbps.

            The 100Gbps stuff is pushing the theoretical limits of copper and will likely only be for very short distances (ie. less than 10m, maybe 3m) but that is a whole other ball of wax that I didn’t bring up.

          • the
          • 7 years ago

          The HDMI cables that use them are more expensive for a given length by several dollars. As length increases, this premium seemingly goes up despite the same active controller being used inside the cable.

            • Deanjo
            • 7 years ago

            They are not that much more expensive over a quality cable. In monoprices case it is just a few dollars more.

            • the
            • 7 years ago

            3 ft: [url=http://www.monoprice.com/Product?c_id=102&cp_id=10240&cs_id=1024008&p_id=3950&seq=1<]$2.57[/url<] vs. [url=http://www.monoprice.com/Product?c_id=102&cp_id=10255&cs_id=1025503&p_id=10206&seq=1<]$10.00[/url<] 6 ft: [url=http://www.monoprice.com/Product?c_id=102&cp_id=10240&cs_id=1024008&p_id=3992&seq=1<]$3.61[/url<] vs. [url=http://www.monoprice.com/Product?c_id=102&cp_id=10255&cs_id=1025503&p_id=9889&seq=1<]$14.05[/url<] 10 ft: [url=http://www.monoprice.com/Product?c_id=102&cp_id=10240&cs_id=1024008&p_id=3993&seq=1<]$5.15[/url<] vs.[url=http://www.monoprice.com/Product?c_id=102&cp_id=10255&cs_id=1025503&p_id=9890&seq=1<]$17.60[/url<] 30 ft: [url=http://www.monoprice.com/Product?c_id=102&cp_id=10240&cs_id=1024005&p_id=3963&seq=1<]$25.97[/url<] vs. [url=http://www.monoprice.com/Product?c_id=102&cp_id=10255&cs_id=1025506&p_id=9430&seq=1<]$40.02[/url<] 50 ft: [url=http://www.monoprice.com/Product?c_id=102&cp_id=10240&cs_id=1024005&p_id=2110&seq=1<]$41.27[/url<] vs. [url=http://www.monoprice.com/Product?c_id=102&cp_id=10255&cs_id=1025507&p_id=9172&seq=1<]$58.16[/url<] Prices are more than a 'few dollars more'. A case for Redmere can be made for longer cable runs (~50 ft) but for shorter cables the price premium above simply isn't worth it.

            • Deanjo
            • 7 years ago

            Redmere is far more then just the length of run. Where they actually shine is on short interconnects. Ultra flexible, light, and low profile make them far more manageable than conventional HDMI cables. Not only are you not having to fight with ferrite cores getting in the way but they also put far less stress on your HDMI ports. Hook a conventional cable up to a small set top box like a Apple TV or Ouya and you will quickly be looking for some double sided tape to prevent it doing a pop a wheely.

            Which would you rather have to run through a home theater centers conduit plugging into a receivers 5+ HDMI inputs? Or which would you rather have to try to lay flat on the back of that wall mounted TV’s inputs?

            [url<]http://s21.postimg.org/d7m606civ/IMG_0824.jpg[/url<]

    • Duct Tape Dude
    • 7 years ago

    [quote<]PC users haven't really embraced the current Thunderbolt implementation[/quote<] That's because of the price premium for consumers, the motherboard trace difficulty for OEMs, the active copper cabling, and limited advantage for peripherals. It's great that TB is advancing, but: -A new connector means limited backwards compatibility (or at least requires yet another adapter) -Competing standards are [i<]still[/i<] "fast enough" for the majority of peripherals (ie: USB 3.0) Ultrabook graphics aside, what do we possibly need an expensive proprietary PCIe implementation for? I still weep for what Light Peak has become.

      • Sargent Duck
      • 7 years ago

      Agreed on all points. Although TB might be superior (espcially being able to power dual 4K monitors on a single cable), it’s price premium puts it a severe disadvantage against USB, considering 99% of actual use will be copying files between devices, for which USB is “fast enough”.

      • Parallax
      • 7 years ago

      [quote<]what do we possibly need an expensive proprietary PCIe implementation for?[/quote<] A new display connector. Growth in HDMI and DP bandwidth has been excruciatingly slow, and are now the main impediment to higher resolution/higher quality displays (even on tablets and e-readers). A 16-lane PCIE3.0 connector can already (theoretically) feed an 8k 120Hz 8bit-per-channel display. This new thunderbolt chip doesn't get us there, but it might get people interested in a higher-bandwidth version (and it puts current video connectors to shame).

        • Helmore
        • 7 years ago

        Then I think you should read the DisplayPort wikipedia page here: [url<]http://en.wikipedia.org/wiki/Displayport[/url<] Displayport 1.3 has a 32.4 Gbit/s transfer speed, is also capable of doing two 4K streams and will be finalized in Q2 of 2014.

          • Parallax
          • 7 years ago

          That’s without the DP interface overhead. Assuming the same overhead percentage as DP1.2, that leaves 25.96 Gb/s for actual data.

          That’s enough bandwidth for:
          4k 24-bit 131 fps
          8k 24-bit 32 fps

          If a lossy compression like 4:2:0 chroma subsampling is used, that would double the effective bandwidth and give 65fps at 8k. I would much rather they put R&D and manufacturing costs of implementing the compression into just making better cables, as I really don’t trust everyone to implement the compression well and I’d rather have a lossless signal anyway.

        • Duct Tape Dude
        • 7 years ago

        I get what you mean, but I don’t think TB is the answer for that (or at least not anymore). Here’s my reasoning:

        -TB uses an x4 link instead of an x16 link.
        -The newest TB can do 40Gbps, but that’s not out yet. This is in the same ballpark as the new DP (as Helmore notes).
        -TB will only be slightly faster than DP. but will require more expensive cables and will be incompatible with any DP hardware without an adapter (this is a new limitation introduced by this version of TB).

        My point is that new technology has to hit price, convenience, and performance targets to become mainstream. TB hits only performance, and it’s not by a very wide margin.
        DP development is slow but at least we’ll be able to rely on it for years to come.

          • Parallax
          • 7 years ago

          TB might not be the answer, but I’m hoping VESA will notice other interfaces catching up in bandwidth and do something larger than a 50% bump in their interface. USB 3.1 is already more than half as fast as DP1.2 (on paper).

    • MadManOriginal
    • 7 years ago

    100W? Mobile device? Sweet, smartphone charging just went from 3 hours to 9 minutes!

      • derFunkenstein
      • 7 years ago

      I believe it’s for charging PCs, rather than phones or tablets.

        • internetsandman
        • 7 years ago

        Agreed. This is the solution that would theoretically let you charge your ultraportable using the same connection that hooks it up to the 4K display, external GPU, and storage that sit at your workstation, this is more for prosumers and the like, rather than normal consumers, but I can’t deny that I wish I had an ultraportable that I could do that with, rather than a separate desktop computer solely for gaming

        If you tried to put 100W through a smartphone or small tablet, it would get a bit toasty, to put it mildly

      • just brew it!
      • 7 years ago

      Don’t bet on it. Without some serious advances in battery tech, attempting to charge your phone in 9 minutes will cause the battery to catch fire.

        • MadManOriginal
        • 7 years ago

        Time to petition for a humor tag.

          • 5150
          • 7 years ago

          As long as we get a “Sarcasm” tag too.

        • Farting Bob
        • 7 years ago

        The Oppo find 7 phone comes with a 4.5 amp charger, compared to the 1.5 amp that many flagship phones use. Obviously it doesnt use 20v to charge it, but batteries can take that load if designed for it.

    • NeelyCam
    • 7 years ago

    [quote<]no ETA for Skylake[/quote<] Sounds like Skylake is coming out by the end of next year: [url<]http://www.bit-tech.net/news/hardware/2014/04/17/intel-skylake-2015/1[/url<]

      • Ninjitsu
      • 7 years ago

      Yeah, supports my find too (see my [url=https://techreport.com/news/26346/report-next-thunderbolt-chip-to-double-bandwidth-add-device-charging?post=817021<]main post under this article)[/url<]

Pin It on Pinterest

Share This