AMD doesn’t sound enthused about Thunderbolt

Much ink has been spilled about those Intel Thunderbolt ports on Apple’s new MacBook Pros, but according to a story by X-bit labs, AMD doesn’t share the enthusiasm. In fact, an anonymous spokesperson quoted by the site suggests AMD doesn’t see much of a point to Thunderbolt at all.

"Existing standards offer remarkable connectivity and together far exceed the 10Gb/s peak bandwidth of Thunderbolt. These solutions meet and exceed the bandwidth utilization of many peripherals," the spokesperson reportedly stated. He (or she) went on to note that DisplayPort 1.2 already supports bandwidth of "up to 17Gb/s," and that total per-channel Thunderbolt bandwidth "is only 20% higher than one PCI Express 3.0 lane and about 52% higher than a single USB 3.0 port."

Now, fair enough: 10Gbps is probably overkill for most devices, and the DisplayPort 1.2 standard indeed supports greater bandwidth, which means Thunderbolt could be a curse rather than a blessing by limiting output to 10Gbps.

My geek sensibilities are slightly offended by some of the numbers quoted, though. A single PCI Express 3.0 lane can push 1GB/s, or 8Gbps, per direction, and USB 3.0 maxes out at 4.8Gbps. That would make Thunderbolt 25% faster than a PCIe 3.0 lane and a whopping 108% faster than USB 3.0—not exactly small potatoes. Of course, with few—if any—Thunderbolt devices available on the market, that point may be moot right now.

Comments closed
    • sjl
    • 9 years ago

    I can see where the 20%/52% came from. 10 Gbps times (100-20)% is 8 Gbps. 10 Gbps times (100-52)% is 4.8 Gbps. It all depends on whether you’re starting from the Thunderbolt speed, or the speed of the other standard.

    That said, the language is definitely wrong. Either Thunderbolt is 108% faster than USB 3.0, or USB 3.0 is 52% slower than Thunderbolt.

    The other things – and I haven’t looked into these in any way – are questions of protocol overhead, and system load. The nice thing about IEEE 1394 (FireWire), compared with USB, for hard drives is that 1394 doesn’t need a CPU to be involved while it transfers data; in fact, it can go from device to device with nothing in between. USB, on the other hand, needs a CPU in there somewhere.

    Also, raw figures don’t necessarily mean much (hence my comment about protocol overhead). The sustained throughput of 1394 (even the original, nominally 400 Mbps) is faster than USB 2 can sustain (nominally 480 Mbps).

    So yeah. Thunderbolt is a nice idea, and I look forward to seeing it used in the real world, but the jury is still out on real world performance.

    • Bensam123
    • 9 years ago

    You can do everything you can do with Thunderbolt with USB3 and other existing specs, I don’t know why people are harping about what great things it will do when it’s just another interface. It isn’t backwards compatible with USB and it doesn’t have widespread already implemented devices ready to plug into it.

    The one scenario where it could actually clear up cord clutter is with monitors, but if you’re going to go that far you could just add such things to a keyboard which doesn’t need display data. Especially more so when USB3 finally starts to hit keyboards and it can function as a hub with ACTUAL amounts of bandwidth.

    This is just like another gimmicky Firewire deal and Apple and Intel will try to keep it alive far, far after it’s already dead.

    • mutarasector
    • 9 years ago

    This view seems a tad shortsighted if you ask me considering Thunderbolt/Light Peak bandwidth can scale up to 100Gbps, and doesn’t suffer from potential line noise/interferance of copper. This view must be specific to the introductory v1.0 Thunderbolt ‘copper peak’.

    It certainly doesn’t make sense to jump on it *now* while it’s still a nascent I/O technology. Besides, Intel has already said Apple will have it for a year first anyway.
    but as Thunderbolt evolves and scales up in bandwidth, I think the potential simply can’t be dismissed by today’s I/O standards.

    Thunderbolt and USB 3.0 will be complimentary I/O technologies for a few years to come.

      • Sahrin
      • 9 years ago

      This is a pretty silly position. USB has been extended 3 times now (1.1, 2,0 and 3.0). We’ve already established that TB is going to be multimodal (current implementations are copper, there will be a mixed implementation over copper/optical in gen 2, and then in gen 3 we will get optical only). There is no reason this couldn’t’ve have been done inside the USB spec (it’s funnier when you consider that Intel *designs* the USB spec as well – meaning that Intel had a secret team inside the company that was designing a proprietary interconnect – I wonder what instructions were given to the USB team to make sure they didn’t compete with TB?).

      Everything in TB could’ve been included in USB3 or a later variant in some form or fashion.

        • sweatshopking
        • 9 years ago

        comes back to that greed thing you love so much.

        • NeelyCam
        • 9 years ago

        [quote<]There is no reason this couldn't've have been done inside the USB spec (it's funnier when you consider that Intel *designs* the USB spec as well - meaning that Intel had a secret team inside the company that was designing a proprietary interconnect - I wonder what instructions were given to the USB team to make sure they didn't compete with TB?). [/quote<] And you don't think it's at all possible that Intel wanted USB3.0 to be faster, others pushed back saying it's too hard and doesn't make sense, and Intel said "Screw you guys, I'm going home" and went to develop ThunderBolt?

          • Sahrin
          • 9 years ago

          No, I don’t. There is not yet a single application proposed for TB that is inadequately served by USB3 or SATA or DP. In addition, TB is coming some 3 years *after* the original USB3 spec. If Intel had invested those resources in creating a free-to-the-industry USB4 spec we wouldn’t be having this conversation. They’re willing to do USB3 for free, and despite what you may think it’s a non-trivial technical application (backwards compatibility *plus* high speed). The TB spec is *designed* in such a way as to keep the existing interconnects running through adapters (meaning the only real ‘advantage’ of TB – universality – won’t exist for at least a generation of the tech, which in peripherial terms is probably more than 5 years).

          The *only* thing that will make TB more wide used than USB flavors is if Intel delays design of the UHCI, then holds it out of their chipsets, then pushes TB through upmarket vendors first.

          Oh, wait.

          I really have no idea why you sheeple (and I’m mostly talking to the Tech Press) are acting like this is some benign thing. Intel is murdering an existing standard for no apparent reason. Even if you assume complete altruism on their part (ie they get no royalties – which is asking a lot to believe given Intel’s history of bribery and anti-trust violations), Intel is still destroying every peripherial owned today (by adding a cost to it – either wholesale replacement or adding a dongle). And for what? Do you honestly believe that a 10Gbps multi-channel controller chip over copper is cheaper than various integrated controllers on the south bridge? What’s the advantage? The only thing that makes sense is Intel is trying to make another interconnect proprietary. First the FSB with QPI which murdered the chipset market, now they want the controller market (made up of companies like Via and Realtek). What’s next?

          I wish to God the FTC and DoJ would get off their fat regulator posteriors and ram a federal investigation wake-up call up Intel’s asses. Is it impossible that Intel’s plan is different then I’ve laid out? Hell yes it is, but when you’re a convicted market abuser you lose the benefit of the doubt. It’s only AMD’s goodwill (and desperate financial situation) that has saved Intel a trip to federal court for possible breakup. They’re already starting the same anti-market shit. It hurts consumers, it hurts the market, it hurts the progress of technology. And for what? A couple hundred million a year in revenue from the controller chip market?

          Intel on-died the northbridge, and motherboard prices collapsed. No, wait. That’s not right. They sold half the chipset for the same money. And wiped the chipset divisions out of half a dozen companies, plus their manufacturing base.

          And they can’t even build a stable SATA controller, to boot.

          /rant Sorry – but I’m sick of hearing the Intel line here. If you’re getting geeked out over TB, you need your head examined. It’s a *peripherial* interconnect. It’d be like salivating over PEG, and we all know how much a boost *that* was to graphics performance /sarcasm.

            • NeelyCam
            • 9 years ago

            My god, there is so much stuff so wrong with your anti-intel rant, but I’m tired, I gotta get up early tomorrow, and trying to discuss things logically with you is like ramming your head against a wall. So I’ll pass.

            • Sahrin
            • 9 years ago

            Keep drinking the Kool-aid.

            • NeelyCam
            • 9 years ago

            See reply below.

            • mutarasector
            • 9 years ago

            My thought exactly. Starting with his assertion that TB could be done inside of the USB 3.0 spec.

            Obviously he isn’t aware that Intel has said it currently requires direct access to the PCIe architecture, can’t be implemented even as a PCIe add-in board, or that future scaled up iterations could possibly even bypass the PCH controller itself.

            • NeelyCam
            • 9 years ago

            [quote<]In addition, TB is coming some 3 years *after* the original USB3 spec.[/quote<] And it's significantly faster, much like a USB4.0 would be. Except that it's implemented NOW instead of later. [quote<]backwards compatibility...[/quote<] Backwards compatibility ALWAYS comes with some compliance issues, but all these just slow down the link or adds extra crap on it that doesn't need to be there. A new, clean link can operate faster without all this legacy crap. [quote<]The TB spec is *designed* in such a way as to keep the existing interconnects running through adapters (meaning the only real 'advantage' of TB - universality - won't exist for at least a generation of the tech, which in peripherial terms is probably more than 5 years)[/quote<] The adaptability is a bonus that TB links will have (including the ability to support optical links for long distances)l The main feature for now is the ability to run at faster data rates... There is no comparable I/O link standard as of today. [quote<]Intel is still destroying every peripherial owned today[/quote<] You mean, by offering a faster, more capable option? I thought that was the whole point of competition.. [quote<]And they can't even build a stable SATA controller, to boot.[/quote<] Now that's just pathetic... this was established as a design mistake, and was corrected. How does it relate to [i<]any[/i<] part of the discussion here? I understand that you're tired of hearing the "Intel line", but I'm not so discriminating - if one of these companies offers something substantially different and capable, I'm all ears. I don't knock down ThunderBold just because it came from Intel - I take it for its own merits, and to me it looks like a capable I/O link. Whatever bitchfest you want run here is your call, but I'm free to ignore it all.

    • Krogoth
    • 9 years ago

    Thunberbolt = Firewire redux

    Enough said.

    • Sahrin
    • 9 years ago

    Also, one other thing. Before it gets any further can we get the name changed from Thunderbolt? This will inevitably be initialized into a shorter version, and I don’t know about you but the first thing I think of when I hear “TB” is not a computer interconnect.

    • OneArmedScissor
    • 9 years ago

    People are throwing around a lot of, “But what if you could use Thunderbolt for X in the future?” Wrong question.

    What if none of this mattered, right now, and Intel themselves already proved it?

    In a day and age where Intel can put 1080p wireless transmitters in laptops, and reduce “chipsets” to one southbridge, instead, they’re pushing more proprietary connectors, adapters, extension cables, and controller chips…under the guise of unifying everything, reducing manufacturing costs, and saving power?!? Wut?!?!?!/1/1?!?!?!

    My mouse was wireless 6 years ago. Darn near anything can be wireless at this point. This isn’t about trying to look past current shortcomings to the full potential of what’s to come. It’s about what we already have.

    And yet, here are some technologically inclinced people, who probably also had a wireless mouse 6+ years ago, talking up this mythical “single connection” that involves more junk to daisy chain together, more junk inside the computer, and more junk you have to pay for, just to do, at best, the same things.

    Really, that’s your definition of progress? *Brain asplode*

    • PRIME1
    • 9 years ago

    Things AMD is not enthused about:
    [list<] [*<]Thunderbolt [/*<][*<]Physx [/*<][*<]CUDA [/*<][*<]Market share [/*<][*<]1st place[/*<] [/list<] Things AMD is enthused about: [list<] [*<]Hotfixes [/*<][*<]Firing CEOs [/*<][*<]Ignoring the mobile market [/*<][*<]Pointing out other companies mistakes with snarky Valentine's day cards. [/*<][*<]Paper launches[/*<] [/list<]

      • NeelyCam
      • 9 years ago

      I’m glad you’re back 100% – I was getting worried.

      • destroy.all.monsters
      • 9 years ago

      Ouch.

    • JMccovery
    • 9 years ago

    I just don’t see the point of piping DisplayPort data through Thunderbolt. Is DP bi-directional? Does Thunderbolt support the AUX channel from DP? External PCI-E cables make more sense, they have higher bandwidth and stream video over the PCI-E connection.

    For NAS, isn’t a GbE port enough, and if a NAS isn’t connected by Ethernet, is it no longer ‘Network Attached Storage’?

    I’ll agree that Thunderbolt is a solution to a problem that currently doesn’t exist. Maybe if Intel introduced Thunderbolt as a replacement for external PCI-E cabling (that had more than 10Gbps bandwidth), it would go over better.

      • thesmileman
      • 9 years ago

      Mr. Jobs only wanted one cable and intel wanted strong initial support that is the reason for DisplayPort over Thunderbolt. I am also sure that is why they went to cable so they don’t need power cables also.

      • bthylafh
      • 9 years ago

      Remember those old Apple Cinema Displays that had just the one cable between them and the computer? It handled DVI and USB and power. You can’t run a monitor on 10 watts, but you might want to run USB and audio along with the video signal.

      • stdRaichu
      • 9 years ago

      The problem is that 10Gb/s isn’t much when it comes to running stuff like PCIe; that “only” gives you 1.2GB/s in any given direction – compare to a single PCIe 1.0 lane at 250MB/s or 2.0 at 500MB/s. And you need to have a PCIe-based controller at each end, since it can’t shunt random data packets, which limits its effectiveness for generic protocols, which will all need dedicated controller support if I’m understanding the technology correctly.

      DisplayPort uses bi-directional signalling for the aux channel, but not for the video data which eats the >9Gb/s. If thunderbolt encapsulates the DP protocol then logically that should include the aux channel as well.

      1Gb ethernet might be fine for most people (heck, most people are fine with sub-54Mbps networks cos all they ever do is farmville) but those of us who do video work would love to be able to plug their computer straight into their “NAS” to turn it into a SAN of sorts. My NAS (QNAP TS 659+) has six discs in RAID10 and it’s limited by 2x gigabit bandwidth, even if you use LACP on your network switch, and the problem goes even more so the more spindles you have in your storage; maxing out a gigabit network connection isn’t uncommon. eSATA can’t be used to hook a desktop straight into a NAS AFAIK, so it would be nice to have something like lightpeak as a poor man’s fibre channel/10GbE. Something that might particularly appeal to the traditional heavy users of macs, audio/video professionals.

      I can’t really see the problem it’s meant to solve either. Plug laptop into TV to get audio and video? Doable with HDMI/DP already. Dock laptop into docking station with dedicated storage (why?) and proper keyboard/mouse/display? Already more easily achieved with a docking station that supports a generic interface (e.g. USB).

      • retnuh
      • 9 years ago

      Thunderbolt is an external PCIe connection, its just a transport layer and more or less a PCIe x4 external plug. And yes the AUX channel on DP is bi-directional. Thunderbolt is actually a dual 10Gbps cable with DP using one of those channels.

      1 GbE only gives you ~120MB/s peak, you really want to bond multiple connections to get every last bit of speed out of an external array. SATA II is 3Gbps and SSDs are already saturating that, so no GbE isn’t enough looking forward and 10GbE is too costly plus adds a bunch of unneeded protocol layers.

      Faster versions will happen over time, this was originally Light Peak as in a optical connector but the costs/practicality of consumer optical isn’t there yet, when/if they switch think upwards of 100Gbps, but I’m not sure if thats per cable or channel.

      • demani
      • 9 years ago

      You can have a fiber NAS (after all fiber is just the conenction- you can run other protocols like TCP/IP over it easily).

      And tell any video editor who is working on location that they don’t have a problem not being able to connect to fast storage. See what reaction they give you.

      Is it niche? sure. But it’s not a solution in search of a problem.

    • End User
    • 9 years ago

    As a Mac user I am looking forward to Thunderbolt appearing on the the next rev of the MacBook Air. As someone who builds his own PCs I am hopeful that Thunderbolt will be available in standalone PCIe cards but “the Intel Thunderbolt Technology Brief does not give a conclusive answer if a Thunderbolt implementation can be realized without a video source”.

    [url<]http://goo.gl/bBIfH[/url<]

    • Ricardo Dawkins
    • 9 years ago

    “I find it very hard to believe that Apple would have pushed so hard for this if they didn’t have big and specific plans for it.”
    of course, their big plan is remove every single port on your Mac, add just 2 (maybe 3) thunderbolts ports and make your pay money for extra accessories if you need connectivity for your devices with Ethernet, USB, FireWire, eSATA jacks.

    That is their master plan and the suckers that pay $2000 for a computer with standard hardware are going to fall for it.

      • tay
      • 9 years ago

      Considering Thunderbolt connectors aren’t Apple specific you are free to buy connectors not from apple. Unlike the iphone/ipod connectors that Apple charges a licensing fee for if you buy non-Apple connectors. Of course the non-Apple connectors are cheaper so I buy those anyway.

      • NeelyCam
      • 9 years ago

      It’s a free world – you can choose to buy whatever laptop you wish. Those who pay $2000 for a MacBook Pro must see some extra value in it compared to standard Windows lappies.

        • charles_chang
        • 9 years ago

        i asked a coworker why she wanted a mac instead of a PC
        turns out it’s because the number of choices of PCs hurt her brain.

          • JustAnEngineer
          • 9 years ago

          This is a very common complaint.

          • OneArmedScissor
          • 9 years ago

          It’s not that much more complicated when you break down the product lines. For example, Apple has Macbooks and Macbook Pros, and Dell has Inspirons and XPSs.

          However, a Macbook is just a Macbook, and a Macbook Pro is just a Macbook that’s…Pro. Everyone can remember that. The problem on the PC side is a complete void of identifiability.

          I can’t name a single mainstream laptop model off hand, from any company, because all I ever see advertised are completely nonsensical labeling schemes like BJ69Pr0nM31st3r9001x. Sound badass? Yeah, well, that’s the new 9″ Acer netbook.

          I just looked at the HP site and they have a “HP Compaq Presario CQ56Z series.” What…the…FUUUUUUUUUUUUUU

          • clone
          • 9 years ago

          that’s a solid example of how consumers view computers in general with far less passion than pc makers would like.

          you don’t see ppl going to the “food” store to buy “food” in cans labeled as “food” only.

          but you do see that with computers because ppl don’t want to be forced to care to the majority they are tools nothing more despite how complicated and important they are.

          you do see ppl going to grocery stores to buy soft taco shells along with salsa, shredded chipotle cheese, sour cream, ground beef with tex mex taco seasoning so that they can make taco’s for dinner… as an example.

    • bcronce
    • 9 years ago

    Of course this initial release version isn’t as awesome as future versions. The current one uses copper, the future versions will be optical. Also, “thunderbolt” will scale to 40gb and 100gb over the next 15 years.

    Some people say it’s overkill, but I say “bring it on”. I would love to have one connector for everything. It will drop down prices and make it so you don’t need to worry about juggling several types of connectors.

    Who can actually say they would rather have 5 types of cables instead of 1.

      • Squeazle
      • 9 years ago

      So… nothing else is going to scale up over the next 15 years?

        • bcronce
        • 9 years ago

        So, your argument against TB is why develop 1 technology when you can waste time and re-create the wheel another 4 times?

        Yes, lets make sure we never have a standard and lots of different cables. I say we ditch USB and make another 10 connection techs because life was so much better back with PS/2, comm ports, LPT ports, and serial ports.

        • mutarasector
        • 9 years ago

        No copper based standard will scale up to what TB will when it goes fully optic. RF interference, noise that requires more conductor pairs for differential signalling, etc. Optic transmission simply sidesteps these issues.

          • NeelyCam
          • 9 years ago

          The key issue optical fiber addresses is the high signal loss in copper wires at high frequencies. Short electrical links could be possible at very high data rates, but a 3m copper cable wouldn’t work (maybe 1m). I bet TB is already differential.. that kind of data rate would be very difficult to achieve otherwise.

      • WaltC
      • 9 years ago

      It’s not overkill at all–quite the contrary, it is exceedingly anemic and underpowered if your goal is to “have one connector for everything”….;) For instance–you can read my post elsewhere in this thread–but do you know why we have multiple PCI/e lanes? It’s because one lane, or two, is just not enough–and we aren’t talking about anything close to “everything”…;)

      As far as predictions about what a proprietary technology will do in the next *15 years*–you need to run like hell away from such technology and don’t look back…;) Hint: Intel let it slip that the P4 was going to ramp up to 10GHz over the course of several years. The reality was that the P4 architecture didn’t make it to half of that before it was cancelled by Intel.

      You will never, ever have “one connector for everything”–because that would really, really suck, and throw general computer bus technology back 15-20 years.

      I wanted to add this, because in a sense we already have “one connector for everything” : we call it a “motherboard”…;)

        • demani
        • 9 years ago

        WaltC wrote”It’s not overkill at all–quite the contrary, it is exceedingly anemic and underpowered if your goal is to “have one connector for everything”….;) For instance–you can read my post elsewhere in this thread–but do you know why we have multiple PCI/e lanes? It’s because one lane, or two, is just not enough–and we aren’t talking about anything close to “everything”…;)”

        Wait- what external bus (since that is what this is about) does Thunderbolt not beat? FC only goes to 8GB, and 10Gb Ethernet is incredibly expensive still. DisplayPort? Well this is Displayport too…
        Outside of Graphics cards and dedicated RAID cards there aren’t any other uses that multiple PCIe lanes are needed for. And everything else is still on the motherboard.

        So what is it anemic compared to? Compared to an internal bus? It’s a non-argument-the two aren’t related. So it might not be the fastest for video-it doesn’t need to be (the USB monitors and video cards have been doing fine). And there will be future versions as well that will scale even higher. But since most laptops would already support a second monitor, we are really just talking about adding a third monitor to a laptop, and then wanting that to be higher performing than the internal one. Seems like an odd request with a remarkably niche application (how is Matrox doing again?).

        This won’t ever supersede all other ports, and anyone who thinks so is clearly smoking crack- that’s why there are other ports still on the MacBook Pros. USB isn’t going anywhere, and if Intel ever gets around to integrating it (I guess next year will finally be the year) then it will be on every machine. But it is hardly anemic-that is a lot of external bandwidth to work with-no laptop has a faster external port (nor do 99% of desktops). It is definitely a step forward.

          • WaltC
          • 9 years ago

          [quote<]Wait- what external bus (since that is what this is about) does Thunderbolt not beat? FC only goes to 8GB, and 10Gb Ethernet is incredibly expensive still. DisplayPort? Well this is Displayport too... Outside of Graphics cards and dedicated RAID cards there aren't any other uses that multiple PCIe lanes are needed for. And everything else is still on the motherboard.[/quote<] Doesn't matter if Thunderbolt is the fastest peripheral bus on the planet--it's still a peripheral bus. It doesn't have the design or the bandwidth to be able to handle external PCIe graphics *along with everything else* people imagine for it--"one cable for everything." I really don't understand the "outside of graphics cards and so on" mentality. The original premise here, and the one to which I objected, was the notion that you could take a MacBook or an iPad and connect them to some kind of as-of-yet unreleased monitor, or even multiple displays, with a built-in PCIe graphics slot to accommodate a full PCIe graphics/3d card, as well as a web cam, USB, etc. ad infinitum--and you could do all of this through a single Thunderbolt port. My point was simply that that was a lot of bandwidth and a lot of devices for the MacBook/iPad cpu to handle--Thunderbolt port notwithstanding. Now, the port itself, as a peripherals port, is surely not anemic--but when paired with possibly anemic MacBook cpus and definitely anemic iPad cpus the whole picture becomes problematic, in the same way that pairing USB with a slow cpu directly diminishes the performance of the USB throughput. Basically, I have no arguments against Thunderbolt as a peripherals port; my arguments are against this "one connector to rule them all" notion in regards to a MacBook or an iPad, specifically. As was pointed out--even USB can handle certain things provided the cpu horsepower is there in abundance--so obviously Thunderbolt can, too. But when you throw the external PCIe 3d-card graphics bandwidth in there, too, as was originally suggested, that's when even Thunderbolt fails to meet expectations. I have nothing against the bus, certainly. But it seems certain that it is not going to meet some of the expectations that I've read for it in this thread. People also had a lot of expectations about PCIe x16 3d-card performance, as well, but limiting factors, including the fact that most 3d-cards boast internal texturing bandwidths multiples of times faster than PCIe x16, that there really is no practical difference between a discrete 3d card running on PCIex8 and PCIex16. I think that Thunderbolt will see its share of bus and peripheral-related overheads, too. It will be interesting to see how the hardware tests out.

            • demani
            • 9 years ago

            Wait- so one 10Gb channel wouldn’t be enough to run *some* graphics card? Nvidia’s Quaddro Plex runs dual 5800s off a 8x slot. Seems like the math would suggest a single 5800 could be run off a 4x if needed. And if the laptop is shipping with integrated graphics then it would certainly be a win (AMD had shipped a laptop ExpressCard video card with a 5670 being the max over that connection, and that was just fine for a second or third screen, and even as a primary compared to the integrated intel graphics on many laptops). For a thin and light, the bandwidth is still more than enough to be a benefit should someone want to use it that way-those aren’t known for having anything too solid as far as graphics cards go. Same for tablets. It doesn’t have to be a Quadro 5800 to be a net win. And BTW- remember that the CPU in those things are currently handling all of those other devices already-the video card is the only new addition.

            I don’t disagree with your overall point that Thunderbolt isn’t going to be the one and only cable for a desktop computer (certainly not, ever), or even for a laptop of the thin and light variety (USB 2 will still have its place for quite a while, and Ethernet is too handy). Monitors will need to surpass 2560 × 1600 before one link at the current spec isn’t sufficient-so it’s plenty for the next two years. Run two lanes, one for video to a 30in display, and one for USB3, Gigabit Ethernet, and eSATA, and you’d still have some leftover. What other ports do you have on your laptop (and would they ever be pegged at once)? The video card thing- it could be useful, particularly with wider channels or a second port. But it still has potential to be way more useful. Personally, I think the market will push in the direction of giving a laptop access to a GPU accelerator (Fermi and whatever AMD is cooking). Take a laptop, and add in some secondary horsepower and the potential to have a solid desktop-class workstation exists pretty quickly.

            Personally, I would love to see some sort of local SAN using it as the interface. Cheap enough for small workgroups to implement it, but fast enough to move chunks of data around (or to have a home SAN).

            • NeelyCam
            • 9 years ago

            Just to double-check: you do realize that the amount of data passing between a CPU and a graphics chip is nowhere near as massive as the amount of data passing between a graphics chip and its dedicated GDDR memory..?

            I assume you know this, but some of your comments make me wonder..

    • bthylafh
    • 9 years ago

    So, is Light Peak proprietary or not? It sounds not, since the unspoken subtext is that it would be available for AMD to use, should they want to.

    This leaves us with:
    1) There’s a royalty, and AMD thinks it’s too high, or
    2) AMD got caught with their pants down and haven’t got a LP controller ready, and there’s no non-Intel 3rd-party chips yet either.

    I’m going with #2 myself.

      • Goty
      • 9 years ago

      Light Peak wasn’t developed by a consortium in the way that USB specs are developed, so of course AMD doesn’t have a controller available; Intel is the only one that has it and that’s the way they like it. Chances are they aren’t even able to legally make a light peak controller and must either buy them directly from Intel or license the technology from them.

      Regardless of the details, this interface is going to go the same way as Firewire. Sure, it looks good on paper and may perform better in the real world, but USB is cheaper to implement and people know what it is.

      • stone2020
      • 9 years ago

      I’m going with #3 myself

      3) Intel dragged its feet on USB3 because they were having trouble with Thunderbolt (say it with a deep voice).

    • Corrado
    • 9 years ago

    Everyone pissing and whining about USB3, the ONLY thing USB3 is useful for right now is external storage. Not even the flashdrives you buy locally are going to be any faster because they aren’t even taxing USB2. You have to SPECIFICALLY buy an SD card that supports anything over 20mb/sec.

    I have absolutely ZERO use for USB3, BUT, how nice is something like thunderbolt on the MacBooks where I can plug 1 single cable from my laptop to my monitor and have a USB, FireWire, Webcam, Mic, and external graphics card in the monitor? Imagine your MONITOR having the PCI-e slot in it, and your laptop has whatever it has now, so you can not need to upgrade your laptop when the CPU is fine but the GPU is lackluster? Can USB3 do that? I think you know the answer.

      • tay
      • 9 years ago

      Yeah thunderbolt is great. Awful name, good idea.

        • Corrado
        • 9 years ago

        I agree that the name is terrible.

          • mutarasector
          • 9 years ago

          More than likely, ‘Thunderbolt’ won’t be the name for future iterations of what has formerly been code named ‘Light Peak’. Maybe ‘Lightning Bolt’ when it goes completely optic? I haven’t checked on this, but who is to say someone else couldn’t simply rename their own LP implementation something different much like Sony did when it opted not to go with ‘Firewire’, but ‘i.Link’ instead?

            • TO11MTM
            • 9 years ago

            Sure. Intel can get this guy to market it: [url<]http://www.youtube.com/watch?v=j_ekugPKqFw[/url<]

      • HisDivineShadow
      • 9 years ago

      USB3 does add an increased power spec so that it can power larger 2.5″ drives more easily than USB2, which often has to fall back to external power or multiple USB2 ports to make up the difference.

      • shank15217
      • 9 years ago

      So to use these devices I have to buy a media hub of sorts, doesn’t that make things even more expensive and bulky?

      • passive
      • 9 years ago

      What are you talking about? Who’s selling, or announcing, a monitor with a built-in graphics card?

      Besides that, sure, it’s nice to have all those other things in the monitor, and just have to plug in one cable. But there are already monitors like that which have USB support. Adding ThunderBolt will drive up the cost of these monitors, and given that for the next few years they will be useful to 0.1% of the population, they will probably carry a substantial premium.

      On the other hand, there are also monitors that support display over USB, so that could be a 1 cable solution as well.

        • Corrado
        • 9 years ago

        This is POSSIBLE with Thunderbolt, not with USB. That was my point.

          • passive
          • 9 years ago

          Why not? There have been USB monitors for 4 years now, and even on USB 2.0 they could do 1600 * 1200 (using lossless compression).
          [url<]http://www.xbitlabs.com/articles/other/display/samsung-sm940ux.html[/url<] There are certainly some issues with it, like a lack of OS support, but I'm sure OS support is required for ThunderBolt as well.

            • Corrado
            • 9 years ago

            But they’ve all had trouble with video stuttering, etc. They’re fine for 2D work, not for anything else. You couldn’t feed an actual GPU over USB3. Being that TB is PCI-e external, it easily could.

            • Sahrin
            • 9 years ago

            With a new special controller on the end. If you give USB an extra controller, you do on-the-fly compression and are able to do the same thing.

      • thesmileman
      • 9 years ago

      USB3 is useful to me. I have 6 ps3 eyes running 640×480 at 60fps uncompressed. Without it I can only run one camera at that speed because USB2 is so slow.

      • WaltC
      • 9 years ago

      [quote<]I have absolutely ZERO use for USB3, BUT, how nice is something like thunderbolt on the MacBooks where I can plug 1 single cable from my laptop to my monitor and have a USB, FireWire, Webcam, Mic, and external graphics card in the monitor? Imagine your MONITOR having the PCI-e slot in it, and your laptop has whatever it has now, so you can not need to upgrade your laptop when the CPU is fine but the GPU is lackluster? Can USB3 do that? I think you know the answer.[/quote<] This sounds far more like an existing dock for an extremely anemic laptop than anything else...;) Or, imagine a *desktop computer*, if you will, in which you can upgrade your components individually when you need to! What a concept, right? Aside from the fact that your idea is not new and has already been done in a myriad of different ways over the years, I have two words for you to consider: Bus Contention. There is a valid reason why we have separate buses for separate devices inside a computer system, and bus contention is the primary reason. Imagine that you have USB, FireWire, Webcam, Mic, and an external graphics card all vying for bus priority on a single two-channel Thunderbolt Port--it will be wreck city and it will be slo-o-o-o-o-o-ow. In short, the use for Thunderbolt that you suggest will not be the use that a Thunderbolt port will serve--and precisely for that reason. There are very, very good reasons why computer technology has evolved as it has over the decades--and for the multiple devices we connect to our computers, One Port is never enough...;) A little perspective on Thunderbolt bandwidth. The internal memory bandwidth in my HD 5770 3d card is 76.8 gigabytes per second. And with its 128-bit memory bus, the 5770 is considered fairly slow in that department compared to its 256-bit bus big brothers. As far as raw bandwidth numbers go, Thunderbolt is 100% in the Ho-Hum-Yet-Another-Proprietary-Bus category.

        • NeelyCam
        • 9 years ago

        [quote<]A little perspective on Thunderbolt bandwidth. The internal memory bandwidth in my HD 5770 3d card is 76.8 gigabytes per second. And with its 128-bit memory bus, the 5770 is considered fairly slow in that department compared to its 256-bit bus big brothers. As far as raw bandwidth numbers go, Thunderbolt is 100% in the Ho-Hum-Yet-Another-Proprietary-Bus category.[/quote<] Bad comparison. Thunderbolt isn't meant to replace internal memory buses. That's what this is for: [url<]http://www.micron.com/innovations/hmc.html[/url<] But I would agree that Thunderbolt doesn't quite have enough bandwidth to feed a medium-to-high-end graphics card as a PCIe replacement - it could become a bottleneck. Maybe next gen (20Gbps? 40Gbps?) could be sufficient...

          • Sahrin
          • 9 years ago

          [quote<]Bad comparison. Thunderbolt isn't meant to replace internal memory buses. That's what this is for: [/quote<] ...hehehe. You realize that Micron says that the thing you linked to is ... internal memory (DRAM)?

            • NeelyCam
            • 9 years ago

            Sure I do – I gave that as an example to WaltC because he was talking about internal memory and Thunderbolt together.

            • NeelyCam
            • 9 years ago

            [url<]http://news.cnet.com/8301-13924_3-20031380-64.html[/url<] : [quote<]The secret sauce is the memory controller (see graphic) that's been added to the memory. "By putting this logic layer--which is actually a controller chip--we were able to overcome that bottleneck by crafting a higher speed bus that will go from the [controller] chip to the CPU. A very, very high-speed bus," he said.[/quote<] Sounds a lot like a Thunderbolt (or many in parallel), now doesn't it..?

            • Sahrin
            • 9 years ago

            Not really, considering LP was announced 2 years ago.

            • NeelyCam
            • 9 years ago

            What does that have to do with anything?]

            The point is that the industry is moving to a higher data rate per lane – that’s what TB and this Micron thingy are implying. And the current standards such as USB2/3 and SATA6 are having trouble catching up

          • WaltC
          • 9 years ago

          [quote<]Bad comparison. Thunderbolt isn't meant to replace internal memory buses. [/quote<] Understood. I only used that example to make the point that two-channel Thunderbolt is not such great shakes in terms of the raw bandwidth deployed elsewhere inside computer systems for many things. [quote<]But I would agree that Thunderbolt doesn't quite have enough bandwidth to feed a medium-to-high-end graphics card as a PCIe replacement - it could become a bottleneck. Maybe next gen (20Gbps? 40Gbps?) could be sufficient...[/quote<] Yes, and if it cannot even do that, how might it ever try and do that and also do USB, FireWire, Webcam, Mic, etc., *at the same time*? It simply won't work. It isn't just the bandwidth, either. We've got separate buses inside x86 standardized tech for many things so that lots of things can run asynchronously and so that many buses can run at the same time. Even if you had the bandwidth required, use of a single Thunderbolt bus which deploys bandwidth to its attached devices in a serial fashion is going to run into performance problems when the devices on the bus all start demanding immediate access to the cpu. Multiple buses in x86 tech greatly alleviate such problems. That's why I said that trying to restrict "everything" to a single bus would push us back 15-20 years in cpu and device bus designs. Thunderbolt is likely just fine for its intended use as a peripheral bus--but "one connector to rule them all" it is not...:D

            • Corrado
            • 9 years ago

            1 USB 2.0 port can already do Webcam, mic, and another USB hub. See the Apple Cinema displays.

            • WaltC
            • 9 years ago

            While I won’t argue with your assertion about what USB 2.0 “can do” (since that depends on many things, not the least of which is the power of the cpu you are running), I will only say that my remarks were addressed to the hypothetical “Thunderbolt cable” which supported not only the things you mention, but also an external PCIe graphics/3d card–all at the same time. The things you mention here often consume little to no bandwidth. I was referring to a rather robust use of all of these things, in addition to the bandwidth required for the external GPU. Not only would Thunderbolt prove insufficient, but the iPad cpu is extremely anemic and would bog itself down badly trying to coordinate all of this activity. The Apple Cinema display USB ports are designed around Macs with gpus and cpus far more powerful than what you’d find in an iPad.

            • Corrado
            • 9 years ago

            Who is talking about an iPad?

            If you took an existing MacBook Pro 13″, it has an integrated intel video. Not so great for gaming. But its a quad core sandy bridge. WAY more than enough for gaming. If you have a TB port, and a graphics card in the monitor, you would not be using ‘display port over TB’, you’d be using it expressly as a PCI-e cable. If they can do, say, 4 channels of TB over 1 cable in the future, thats 40gbit/sec. Being that 1 lane of PCI-e 2.0 is 4gbit/sec, it stands to reason that its approximately a 10x pci-e slot. As we’ve seen with multi-gpu setups, an 8x pci-e lane is more than capable of handling a high end video card. This leaves you with at LEAST 8gbit/sec of left over bandwidth for other devices.

            • WaltC
            • 9 years ago

            NeelyCam mentioned the iPad–yes, and certainly a quad-core SB could handle the load.

            [quote<]If they can do, say, 4 channels of TB over 1 cable in the future[/quote<] That's my whole point. Why speculate about the future? Things change so rapidly on the hardware front that it often happens that technology initiatives started years earlier are scrapped entirely because something better comes along. As of the moment, there is no indication that PCIe graphics-card-in-a-monitor expressly for MacBook Thunderbolt is slated for production--which I think is pretty conclusive...;)

            • Corrado
            • 9 years ago

            This much is true, but why standardize on something, when you have something better? If anyone can ‘make it work’ its Apple. They were the first to ditch: Floppies, PS2, Parallel, Serial, among the first to move to LCDs. Apple is the perfect ‘early adopter’ market for things like this because people will buy because thats what Apple customers do, myself included. How long did USB take to really gain a foothold? It was YEARS. I remember having USB ports on my Pentium 200mmx, and no peripherals, and no OS to really use them until Win98 came along. No new technology comes out and is automatically adopted regardless of if its better. AGP took a while to replace PCI. PCI-E took a while to replace PCI (and still hasn’t fully yet). SATA took a while to replace PATA, even though PATA drives at the time came nowhere NEAR the theoretical limitation of the bus. Also, I’m skeptical of the actual throughput of USB3. USB2 has a theoretical max of 480gbit, but in reality its nowhere near that. Firewire 400 peed all over USB2 in terms of throughput speeds, despite the lower theoretical ceiling.

      • FuturePastNow
      • 9 years ago

      The thing about USB 3 is that I can plug all of my old stuff in the new ports and all of my new stuff in the old ports, and it all works.

      • Bensam123
      • 9 years ago

      What is a desktop and/or a docking station? Alan

    • Buzzard44
    • 9 years ago

    I think the person quoted in the article wasn’t very good with numbers. A USB 3 port is 48% of the speed of Thunderbolt, and A PCI-E 3 lane is 80% of the speed of Thunderbolt, thus the person quoted must have used very flawed reasoning to get the 20% (100 – 80) and 52% (100 – 48) figures.

    Also, yeah, I used to think Lightpeak was going to be awesome, but now I’m skeptical. Mainly it depends how far they can push it to me. Right now I don’t have anything that USB 3 wouldn’t be fast enough to push, and USB is much more prevalent than this fledgling tech.

    Maybe in the future, when Lightpeak is faster and I can use higher bandwidth peripherals – then it’ll be nice.

    And what company is ever enthused about it’s opponent’s product?

      • designerfx
      • 9 years ago

      I’d hope nobody is happy about a *proprietary* interface being pushed on devices again. Yes, that’s what lightpeak is.

        • derFunkenstein
        • 9 years ago

        Being proprietary also instantly reduces the size of your market. In the next year or so, this will take off in the Mac market because 100% of new computers will have the port. In the Windows market, probably not so much.

      • jcw122
      • 9 years ago

      They didn’t say what those numbers were: theoretical or practical…so it’s kind of tough to even say they are wrong or right.

        • NeelyCam
        • 9 years ago

        Let’s be fair – the most likely explanation is that marketing folks are generally math challenged.

    • Silus
    • 9 years ago

    AMD not enthusiastic about something their competitor is pushing ? Noooo….that can’t be true 🙂

      • bittermann
      • 9 years ago

      They have a point though…another proprietary standard to push on everybody when the current standards are not being maxed out. Thunderbolt is a solution looking for a problem. Most people want standards that can be adopted by all companies universally me thinks.

        • Silus
        • 9 years ago

        No, they don’t have a point, because they are trying to fight something that’s impossible to fight: that other companies other than them want to make money. And that’s what technology centric companies do. They invent new stuff to sell and make money. It may eventually become a standard, but it’s always something done at a cost.
        AMD is always gloating that they have made so many innovations over the years, but always criticize others for doing the same. Double standards much ? Still that’s normal, which is why I pointed out the obvious: AMD criticizing a technology pushed by one of their competitors ? Say it isn’t so…

        Plus, AMD is looking more and more like Apple. They want people to use the “standards” that they support or where they have invested interests. Again, not surprising for any company, but publicly admitting it, just makes them look bad.

          • WaltC
          • 9 years ago

          I trust you did read the linked article? It goes to xBit, and it involves quotes from an *unidentified* AMD employee. There is no attribution in the article, and no one from AMD is quoted by name in the article–so, while I don’t think this is the case, the entire article could well be a piece of fiction.

          Secondly, assuming the article is indeed accurate in all of the particulars, AMD made no statement here whatsoever in an official capacity. An AMD employee (again, forever unknown) was asked his opinion on a specific topic and provided an answer, apparently (we cannot be sure that this ever occurred however, because the article has no attribution.) That is all that has happened here. AMD has done absolutely nothing to make them “look bad” because they haven’t made any kind of statement on Thunderbolt at all.

          Aside from that–the xBit article linked to here makes no mention of Apple in any capacity–yet in this thread we see Apple people making up all kinds of wild, fantastic fairy tales about “one connector to rule them all,” and so on…:D Poor Intel–can’t catch a break–first they had to cancel Larrabee because all of the irresponsible, third-party hype about the product was creating expectations that Larrabee could never hope to reach–and now we see “Thunderbolt”–which is nothing more than a proprietary bus port added to the myriad of buses already being used inside x86 tech–being hyped all out of proportion by people who still view technology as “magic” instead of just plain technology…;) What will happen should Thunderbolt fail to meet all of these weird and wild speculations? Same as happened to Larrabee–it will simply die on the vine.

          When oh when can we get back to *sane* discussions about computer technology?

            • NeelyCam
            • 9 years ago

            Larrabee was never released. Thunderbolt is released, as part of a new product by arguably the most demanding computer maker on the planet.

            Apples vs. Oranges

            • Silus
            • 9 years ago

            I read AMD spokesperson, so even if he/she remain anonymous, it pertains to AMD nonetheless. But fine, it may not be “totally” by AMD, officially speaking.

            And you’re trying too much to twist/change the article, based on hypotheticals. The person clearly identified himself/herself as someone from AMD, so his/her POV does indeed reflect some (if not all) of AMD’s position on this.
            And also, it would come as no surprise that AMD is indeed against a technology that one of their competitors is pushing, much like it would not surprise anyone that Intel wasn’t too happy about a tech AMD is pushing. it’s just business as usual and that was my point in my very first post.

            • clone
            • 9 years ago

            given it’s slower than current solutions I’d assume it’s blatantly obvious that as a new “feature” it’s kinda weak…… is it bad that AMD craps on a tech that is slower than what’s already available?

            display port 1.2 has higher potential by quite a bit all round, thunderbolt isn’t even remotely interesting.

            • NeelyCam
            • 9 years ago

            Wrong – Thunderbolt is faster. Display Port 1.2 maxes out at about 17Gbps with four lanes. Thunderbolt has 20Gbps in one direction with two lanes.

            • Sahrin
            • 9 years ago

            So far as I know you can’t gang lanes in TB. Making this BS.

            • NeelyCam
            • 9 years ago

            Why not? Would seem like an obvious option..
            Any links?

            • clone
            • 9 years ago

            [quote<]"Existing standards offer remarkable connectivity and together far exceed the 10Gb/s peak bandwidth of Thunderbolt.[/quote<] Thunderbolt is slower by quite a bit, hence it's kinda unremarkable crap atm.

            • NeelyCam
            • 9 years ago

            Was this reply fail?

            • clone
            • 9 years ago

            nope, just reiterating how wrong you are.

            • NeelyCam
            • 9 years ago

            If you’re comparing TB to a multi-lane PCIe, you deserve all the butt-hurt you’re getting.

            Fundamentally, TB is the king of datarates/lane compared to most competitors (there are a fringe few that are significantly more power hungry)

            If you don’t agree, please feel free to list some of your favorite alternatives so we can compare.

            • clone
            • 9 years ago

            their is no “butt hurt” just the reality that you are wrong.

            [quote<]Existing standards offer remarkable connectivity and together far exceed the 10Gb/s peak bandwidth of Thunderbolt.[/quote<] display port 1.2 is faster by quite a bit, I don't care how it's done, that's for geeks to mull over, to obsess over, performance is what matters and display port is currently considerably faster and "far exceeds the 10gb peak of Thunderbolt." no butt hurt at all just the reality that you are wrong. you should go back to talking about how doubling and tripling Thunderbolts potential will really make it faster along with a slew of other things not currently available. it's slower hence you are incorrect.

            • NeelyCam
            • 9 years ago

            No. It’s some 5Gbps per lane in one direction. TB is 10Gbps per lane in both directions. TB is clearly faster.

            Stop spreading fud.

            • demani
            • 9 years ago

            Wait, wait, wait: so we can’t talk about faster TB ports because they have t been released yet, but your counterpoint is that DisplayPort supports bandwidth to support monitors that aren’t released yet? Seriously- find a DisplayPort monitor that needs 17Gb bandwidth. And good luck finding a SAN based on USB, SATA, or DisplayPort. Even 10GB iSCSI is cost prohibitive. TB has the potential to make that possible within a home or small business, since it is way more cost effective. It won’t happen immediately, but it will likely happen soon. And to the thought USB 3 is the solution- remember what happens to a USB port when you plug your mouse in.

            • clone
            • 9 years ago

            this is a double reponse to Neely and demani, FUD…. lol, it’s in the paragraph above just read it.

            [quote<]Existing standards offer remarkable connectivity and together far exceed the 10Gb/s peak bandwidth of Thunderbolt[quote<]the DisplayPort 1.2 standard indeed supports greater bandwidth, which means Thunderbolt could be a curse rather than a blessing by limiting output to 10Gbps.[/quote<][/quote<] as for the other response you can talk about anything you want, Thunderbolt is slower. hence not interesting. as for talk about stuff that doesn't support other stuff, I don't care it's not a part of the discussion, I don't care about USB 3, I don't care about ISCSI.... that devices haven't leap forward to support the fastest affordable existing interface is only a matter of time. Thunderbolt seems to me to be more like a MAC feature like Firewire was, a slower "fast interface" that'll eventually fail and only be solid in the MAC world never gaining significant traction on PC, an afterthought at best.

            • NeelyCam
            • 9 years ago

            [quote<]this is a double reponse to Neely and demani, FUD.... lol, it's in the paragraph above just read it. "Existing standards offer remarkable connectivity and together far exceed the 10Gb/s peak bandwidth of Thunderbolt" "the DisplayPort 1.2 standard indeed supports greater bandwidth, which means Thunderbolt could be a curse rather than a blessing by limiting output to 10Gbps."[/quote<] The problem with the paragraph is that it's incorrect. 10Gbps the peak bandwidth of ONE TB lane. There are two inside the TB cable. That means 20Gbps total in EACH DIRECTION. Displayport needs four lanes to get to that 17Gbps stated here. => TB is faster. This is an indisputable fact. How many times do I have to repeat this before you understand?

          • bittermann
          • 9 years ago

          Yes they have a point (I don’t care if AMD officially said it or not). Please enlighten me on this gloat thing and double standards they have done over the years….what, when, where?

          Like I said this technology is an answer for a problem that doesn’t exist right now….And there is a big difference when AMD pushes OPEN standards that they support vs. some of the proprietary crap Intel/Nvidia pushes….

            • NeelyCam
            • 9 years ago

            [quote<]Like I said this technology is an answer for a problem that doesn't exist right now....[/quote<] So was iPod, iPhone, iPad... Build it and they will come...? [quote<]And there is a big difference when AMD pushes OPEN standards that they support vs. some of the proprietary crap Intel/Nvidia pushes....[/quote<] ? Thunderbolt is faster than USB2.0, 3.0, SATA3/6, PCIe G1/2/3 or even HyperTransport. How does it qualify as "crap"? You almost sound like a pissed off fanboy

            • bittermann
            • 9 years ago

            I like Intel or AMD for whatever fits the best…your the one sounding like a pissed off fanboy?

            WTF…ipad, iphone? Were talking about an interconnect not a cell phone or laptop…ipad and iphone are standards now?

            Maybe you should read up a little more about the differences between Thunderbolt and USB, SATA, PCIE, etc….your talking about a data bus vs. connection standards???????

            • demani
            • 9 years ago

            And with that last sentence you just mixed up data bus and connection standards all into one group…
            The END connectors are connection standards (e.g. L-type SATA or USB-A, mini-USB, USB, 4x PCIe slot). But USB (see how that B stands for “bus”), PCIe and SATA are the data bus portions (though are often conflated to include the connectors as well).

            • NeelyCam
            • 9 years ago

            You didn’t answer my question: how does Thunderbolt qualify as crap?

            iPad/iPhone were given as examples of how completely new and different products were released, ridiculed (remember the maxi-iPad jokes?) and eventually universally embraced. They were addressing needs that nobody even knew they had.

            Maybe I should’ve been more clear on the USB/SATA/etc.: but I was pointing to the transfer speed per lane. Note that all standards I mentioned are I/O standards – for slightly different purposes, but comparable nonetheless.

            • clone
            • 9 years ago

            [quote<]"Existing standards offer remarkable connectivity and together far exceed the 10Gb/s peak bandwidth of Thunderbolt.[/quote<] it's slower than an existing standard hence it's unremarkable crap atm.

            • NeelyCam
            • 9 years ago

            Um, no, unless if you compare to 100Gbps Ethernet..?

            • mutarasector
            • 9 years ago

            re: “They were addressing needs that nobody even knew they had.”

            There is no such thing as addressing needs one is not aware of. Apple did however kick off a trend in mobile computing creating a demand for their products by means of a new user paradigm, but not by simply addressing current needs no one knew of. This is pure marketing mythos rooted in historical revisionism.

            • hapyman
            • 9 years ago

            As do you… you are defending something we know almost nothing about in real world applications.

            Honestly I am not surprised that Apple would commit to a new proprietary port in which they are the only ones currently using. At least for the time being it allows them control over the peripherals and will prob allow them to charge those exorbitant prices we normally see for those dinky cords from apple. HDMI adapter anyone?

            • Silus
            • 9 years ago

            The “gloat thing” was some AMD top guy a while back (probably CEO) claiming that AMD has been leading innovation for years, while Intel did nothing in that time.

            The double standards are obvious. Again, against anything the competition wants to offer, while trying to push whatever they have invested on. Double standards are usual in this kind of business. You can’t promote your product, while also praising your competitor of course.
            The difference between AMD and other companies, in the latest years, is that AMD doesn’t mind throwing sand in the eyes of competitors i.e. playing dirty. Recently the Valentine’s Day Sandy Bridge joke was one. The making fun of Fermi ads are another example. Or even commenting on NVIDIA’s mobile chip failures, based on guesses and nothing else.
            NVIDIA also crossed the line, with the cartoons against Intel, but I didn’t see Intel or NVIDIA making fun of AMD’s Phenom TLB bug.

            AMD supports open standards, because they they have nothing new nor the financial ability to push anything. If they did, they would be doing everything to get their message across, on how their new tech is much better than whatever the competition supports.

    • Sahrin
    • 9 years ago

    Unfortunately, when Intel shoves this proprietary standard down the throats of the market AMD won’t really have any choice but to pay Intel’s exorbitant license fees (assuming Intel licenses the protocol at all). It’s also likely that Intel will force AMD to buy an external chipset while Intel itself includes the protocol in its own southbridges.

    So far, Intel has yet to demostrate anything about Light Peak that couldn’t’ve been integrated into a future USB spec. The cable runs over backwards compatible copper (USB3 does that already), it carries multi-Gbps (USB3 does that already), it carries a massive royalty to Intel for every installation – oh no, wait, USB3 doesn’t have any royalties. I guess it doesn’t have *every* feature Intel wants.

      • bcronce
      • 9 years ago

      Intel said cables and peripherals(Printers/cameras/etc) will be free to license. They plan on that actual chips to have a low licensing fee near $0, but they haven’t said what.

        • Sahrin
        • 9 years ago

        Do you have a source for that?

        And the controller chip and converters aren’t “free.” That’s what Intel is selling.

        $100 is “near” to 0 dollars.

    • FuturePastNow
    • 9 years ago

    I can sort of see the appeal. Tunneling a PCIe x4 link over a cable allows some really awesome peripherals, like a fast RAID external drive or a decent external GPU, or a fully-featured docking station.

    Problem is, those things exist now, and the markets for all of them are very very small. Nobody uses docking stations or external graphics, and anyone who really wants enterprise storage attached to their PC is insane.

    So to attach a normal, every day peripheral like a mouse, flash drive, or printer to a Thunderbolt port, you’ll have to buy a fricking USB dongle to carry around with your laptop. Which no one will do.

      • blastdoor
      • 9 years ago

      It could be that the reason the market for those things is small is because the current implementations are extraordinarily unappealing for consumers.

      A drop-dead simple docking station that connects a MacBook Air to a 27″ monitor that has user-upgradable SSD and HHD storage might be compelling.

      Or, a couple of years down the road, imagine an iPad 4 that connects to that same 27″ monitor, but now the monitor also includes a more powerful GPU than makes sense to include in the iPad itself. When connected to the dock, the iPad could switch from touch-tablet interface to keyboard/mouse-desktop interface.

        • FuturePastNow
        • 9 years ago

        No, the reason the market for those things is small is because very few people are willing to spend money on accessories they don’t absolutely need. That isn’t going to change.

          • blastdoor
          • 9 years ago

          I don’t think think the nice people at Bose would agree with you:

          [url<]http://www.bose.com/controller?url=/shop_online/digital_music_systems/sounddock_systems/sounddock10/index.jsp[/url<]

      • thesmileman
      • 9 years ago

      “So to attach a normal, every day peripheral like a mouse, flash drive, or printer to a Thunderbolt port, you’ll have to buy a fricking USB dongle to carry around with your laptop. Which no one will do.”

      You seem to think USB is going away. It isn’t.

        • NeelyCam
        • 9 years ago

        I could imagine seeing this on an iPad3 or something – all you need to do is connect a Thunderbolt cable, and you can have a full ‘desktop’-like system available… Two large displays, mass storage/optical drives, keyboard/mouse, speakers… while charging up the iPad3 battery.

        I sort of see this as an ultimate docking station replacement.

          • WaltC
          • 9 years ago

          I am surprised at the number of people who ought to know why such an arrangement wouldn’t work, but don’t, evidently. There are so many problems with what you have just loosely and very vaguely described, that it is difficult to know where to begin. I’ll start with bus contention and a very anemic iPad cpu, for starters–then we might get into why a two-channel Thunderbolt port (to get to the “one cable for everything” mantra) doesn’t have close to the required bandwidth for “everything,” etc. There are many more technical things we could discuss that would constitute failure for the “one cable for everything” hypothesis.

          Last but not least, the real solution here is most obvious: you buy an iPad and you buy a desktop–and there you have it…! A superior solution in every way.

            • NeelyCam
            • 9 years ago

            Your condescending attitude doesn’t look pretty on you.

            First, would you please point out why 2x10Gbps each way wouldn’t be enough to do all that?

            Second, you do realize that I mentioned iPad3… iPad2 is probably close to being able to do all this. I would bet iPad3 is going to have four (or more) ARM A9 or A15 cores with some pretty powerful graphics on the side. That shouldn’t be a concern.

            Finally, why would I buy an iPad AND a desktop when I could buy just an iPad?

      • Corrado
      • 9 years ago

      Wrong reply thread.

    • blastdoor
    • 9 years ago

    Thunderbolt is undeniably overkill for today. The obvious retort would run something like this:

    “But such is the “burden” that any new technology must bear. The PS3 was overkill for playing PS2 games. PCI was overkill for existing ISA cards. The Pentium 90 was overkill for running Windows 3.1. etc.”

    And the rejoinder would be:

    “While that is an accurate description of the past, the present and near future are arguably different. In the last five years we have seen an unparalleled overshoot of hardware capability relative to software needs in the PC space. The explosion of netbooks and iDevices, combined with shrinking sales of more powerful PCs, illustrated that the vast majority of consumers are perfectly content with the computing power offered 5+ years ago, and now would prefer greater mobility to greater CPU power.”

    I think that the second is probably the stronger argument… it is much harder to see any obvious need for Thunderbolt. But there’s a huge caveat on that statement: Apple sees a need for it. I find it very hard to believe that Apple would have pushed so hard for this if they didn’t have big and specific plans for it.

    • blastdoor
    • 9 years ago

    “Now, fair enough: 10Gbps is probably overkill for most devices, and the DisplayPort 1.2 standard indeed supports greater bandwidth, which means Thunderbolt could be a curse rather than a blessing by limiting output to 10Gbps.”

    But that’s for a single channel, and at least Apple’s implementation has two channels. The specs for the MBP indicate that it supports the same screen resolutions as with mini-DP, which implies that they certainly are using both channels for the display.

    • jiaolu
    • 9 years ago

    i don’t see it much use of thunderbolt either. why buy the thing you can not use right now. usb3 ecosystem definitely much better

    • dragmor
    • 9 years ago

    A couple of years ago LightPeak would have been great. Basically a PCIe x4 (gen1) for an external mid range graphics card. Now with Sandybridge and Fusion, not so much.

    • rika13
    • 9 years ago

    Of course AMD isn’t enthused, it’s the other guy’s shiny thing. The idea of a true universal connector for anything is damn appealing, no more ethernet, usb, firewire, dvi/dp/hmdi/vga, just a bunch of one port.

      • Vhalidictes
      • 9 years ago

      A universal connector is a good idea… as long as it’s popular. If it’s not, it’s just going to cause problems while it fights other standards.

    • muddybulldog
    • 9 years ago

    The thing about spin is that you count on those hearing it to be left believeing the impression you want to leave, in this case “Thunderbolt is no big deal”.

    But those who actually think about what you’re are saying quickly see the flaws.

    “Existing standards offer remarkable connectivity and together far exceed the 10Gb/s peak bandwidth of Thunderbolt.” — then offers up a single example.

    DisplayPort 1.2 already supports bandwidth of “up to 17Gb/s,” — and that single example is a video/audio interconnect only, so not useful for most computer peripherals.

    total per-channel Thunderbolt bandwidth “is only 20% higher than one PCI Express 3.0 lane and about 52% higher than a single USB 3.0 port.” — and Thunderbolt provides TWO of these channels per port.

      • mcnabney
      • 9 years ago

      The only type of connection that needs that kind of bandwidth (in a non-datacenter environment) is a connection between computer and display which needs to carry an uncompressed video stream.

      What kind of peripheral were you thinking of that needs this connection, since displays are currently handled by Displayport and HDMI?

      Is a massive array of SSDs currently considered a peripheral? That is the only non-display and non-host device I can think of which might actually need even one channel of Thunderbolt. And don’t say ‘external GPU’, since even moderate GPUs would be limited by two channels.

      Sorry Intel, but I think we have the bases pretty well covered.

      USB for low to high speed data, bonus power provided.
      DVI/HDMI basic monitor support
      Displayport – enhanced monitor(s) support
      eSATA/SAS very high bandwidth data

      All Intel is trying to do is wave it’s magic wand and demand that the industry drop multiple open standards and all gather in their propriety tent. Sorry, Intel – We have been burnt by you before.

        • Corrado
        • 9 years ago

        You realize LightPeak supports ALL those standards with a cable adapter? IE, you can buy a hub with a single Thunderbolt plug on one side and USB3, FireWire, DisplayPort and an eSata plug on the other?

        Or that this consolidates it to a single plug, like USB tried/did moving from midi/joystick port, AT keyboard, PS2 keyboard, PS2 mouse, Parallel and DB9 serial ports into 1 ‘universal’ port? They’re trying to move EVERYTHING, display, storage, peripheral, network into 1 single port. If you have 5 thunderbolt ports, and nothing else, you’re still OK because any one of those can plug into anything else.

        Its funny to see people say on their motherboard they want no PS/2, parallel, serial, audio, etc. Just give me 12 USB ports!, they say. But when Intel tries to actually do away with ALL other ports and standardize on a single universal one that would also get rid of the network, and display ports, people are like zOMG WTF? WE’RE FINE WITH THE 5 DIFFERENT PORTS WE ALREADY HAVE!

          • HisDivineShadow
          • 9 years ago

          Last I heard, the dream of one connection, multiple protocols was squashed. You need an adapter (not just a cable adapter) to make Thunderbolt work with USB3, ESATA, or Firewire. The connection (not just the physical connector) is not inherently compatible with those standards.

          That is, a vendor would have to install a hardware adapter for each of those components into the computer attached to the Thunderbolt connection for it to be able to read those connections and only AFTER you also had invested in the connector adapter that is necessary for your peripheral to hook into the T-bolt port.

          All in all, that’s a lot of hassle just to use a USB3 port because Thunderbolt is all you got (left). I agree that the dream Intel sold everyone on of Thunderbolt being used for lots of different kinds of connections across ONE cable was a great one. Too bad, it is a dream they tossed aside in favor of limiting it to only PCIe and Displayport. I’m sure that saved them a lot of coin. How likely is it that vendors will come along that add back the USB3, ESATA, and Firewire compatibility that you want? And how much is that going to cost on top of the already inflated cost of the relatively expensive Thunderbolt adapter?

          USB3 is cheaper, it’s fast enough for 99% peripherals out there, and it’s easily b/c with USB2. Thunderbolt has a place, but imo that place is as a connection for external video cards, RAID arrays, and a step-in for Displaypot when a real port is not available. For the very fastest of SSD’s, it will also replace ESATA in function.

          • PixelArmy
          • 9 years ago

          Well, there’s theory and implementation…

          In it’s Thunderbolt incarnation, LightPeak supports DisplayPort and PCIe. So your cable adapter is a lot more complex than you make it sound, as all these other protocols will need to be “translated” to PCIe.

          Then there’s the bandwidth… Again, they were planning for 100 Gbps. This implementation is 10 Gbps x2 channels for 20 Gbps. So depending on your resolution, if you connect DisplayPort you have 3 Gbps left (20 Gbps – 17 Gbps) which is less than USB 3.0. Now, the saving grace here is the odds that you’ll run that resolution monitor or that you’ll even attach a monitor is lower, but think if this goes to the desktop where you’re guaranteed to run a monitor.

          Of course, look at the MacBooks, they just removed the DisplayPort and added a Thunderbolt port. The fact that all those other ports are still there is says to me, they’re not there yet. Knowing Apple if they could have gotten rid of them, they would have. (Or they could at least be attached via your “single hub” internally).

          I like the idea of LightPeak and look forward to it’s “real” optical form but I’m lukewarm about Thunderbolt. Though I guess you have to start somewhere. And you have to weigh whether a starter implementation will get people used to it, or sour them on it.

          Of course, I could be wrong about all the details.

          • willmore
          • 9 years ago

          So would a simple PCI-E link. If by ‘adapter’ you use a simple PCI-E controller for USB3, SATA, etc.

          So, why not just work on supporting the existing effort to make an external PCI-E spec, intel? You know, the one AMD and others have been working on for years? Oh, right, NIH. Sorry, I forgot you’re a *leader*.

          [url<]https://secure.wikimedia.org/wikipedia/en/wiki/PCI_Express#PCI_Express_External_Cabling[/url<]

        • NeelyCam
        • 9 years ago

        This is the same argument as the “PCIe G2 doesn’t make sense, G1 is more than enough for current applications” or “hard drives can’t saturate a SATA 3Gbps link – this SATA 6Gbps is just a marketing gimmick”

        If there’s bandwidth available, there will be applications making use of it.

        [quote<]USB for low to high speed data, bonus power provided.[/quote<] Thunderbolt provides 10W of power. Then again, you're free to hook up half a dozen cable to your laptop if you insist. I would prefer doing it with a single cable if possible but - hey, that's me.

          • mcnabney
          • 9 years ago

          You are still running lots of cables. They would just be the same kind of cable. And since it only appears like there is going to be ONE port on the MacBook, you are going to need a hub – which is actually going to increase the number of required wires by 1 and the number of boxes on the desk by 1. So a net gain of 2 items in the clutter department.

          And I mentioned the power provided by USB because it is a necessary function for that connection while power is not needed for the others.

          I think that you have this warped idea that you are going to plug one wire into your Macbook and it is going to ‘magically’ connect to four or five items. The best case scenario is that it would potentially allow a single cable to hook to your display, and have the tangle of wires go from the display to the printer, extenal storage…….

            • Corrado
            • 9 years ago

            Right, 1 port… RIGHT NOW. When USB debuted, you got 2. Now theres 12+ available on some boards.

    • JustAnEngineer
    • 9 years ago

    Not Invented Here.

    I believe that Intel has done the industry a disservice by dragging their feet on USB3.

      • stdRaichu
      • 9 years ago

      The wikipedia page doesn’t say much but… is thunderbolt available for third parties? Could AMD implement their own thunderbolt controller, and presumably even if they could they’d need to pay royalties?

      It’s a nice idea, but I’d have much preferred an optical version even though I don’t have a whole lot of use for it myself. I also thing it would have been better to make it a generic interconnect (i.e. no protocol-specific data) rather than restricting it to video + PCIe. PCIe has massive security implications, and I’d rather shunt storage than video. Is light peak switchable? Bundled ethernet + SATA over a single cable to multiple computers would turn the NAS community on its head.

      As an aside: does anyone make a mini display port -> display port cable that doesn’t cost a bloody fortune?

        • bdwilcox
        • 9 years ago

        Is [url=http://www.monoprice.com/products/subdepartment.asp?c_id=102&cp_id=10246<]$7[/url<] too much for a 15' version of that cable you're looking for?

          • stdRaichu
          • 9 years ago

          Not at all, but not seen any in the UK yet.

          And why are they all in white? Does no-one make black cables any more?

        • NeelyCam
        • 9 years ago

        Exactly why would you prefer an optical version? The only benefit is distance, whereas power consumption would likely increase, as would cost (electrical->optical and optical-electrical modules cost money).

          • stdRaichu
          • 9 years ago

          [quote<]Exactly why would you prefer an optical version? The only benefit is distance[/quote<] You just answered your own question 🙂 That said, I've done some reading and I'm not all that enthralled by a "10Gbps" interface is mostly for video data and the thing seems solely designed for hooking up to "all in one" monitors (containing PCIe-attached stuff) for convenience, rather than enabling something cool like fast data access. All a bit meh IMHO.

      • ImSpartacus
      • 9 years ago

      Omg, we’re still hung up by that?

      Idk, I thought it was obvious back when Intel was delaying the USB 3.0 spec. As soon as they announced LightPeak, I figured everybody just understood that they were manipulating the market.

      Long story short, Intel needed more time to perfect LightPeak. To buy time, Intel did everything it could to slow down the adoption of USB 3.0. In the end, it wasn’t enough, Intel couldn’t get it working on fiber and Thunderbolt was born.

      Intel makes good products, but they aren’t quite as angelic as popular opinion seems to make them.

        • NeelyCam
        • 9 years ago

        [quote<]Intel makes good products, but they aren't quite as angelic as popular opinion seems to make them.[/quote<] That's funny - on this site Intel is considered the Devil

Pin It on Pinterest

Share This