A quick look at Thunderbolt on the PC

When Intel’s Thunderbolt interconnect was formally unveiled last year, we learned that Apple would have first dibs on the technology. Device makers would be free to build Thunderbolt-compatible accessories right away, of course, but the first computers to employ the uber-fast interconnect would be Macs.

The enthusiast community met this news mostly with a collective shrug. We already had USB 3.0 ports with oodles of bandwidth for external storage devices that were affordable, backward compatible, and readily available. Thunderbolt hardware? Not so much. We also knew that Light Peak, the precursor to Thunderbolt, was conceived as an optical interconnect. Apple’s implementation ditched optical fibers for copper wires, which simply aren’t as cool.

More than a year after Thunderbolt ports made their Mac debut, PCs are getting their first taste. Select Ivy Bridge notebooks are destined to feature the technology, and Thunderbolt-equipped motherboards are already selling online. A closer look is warranted, so we’ve rounded up a couple of motherboards and an external storage box to see what’s what.

You’ve been… Thunderstruck

Before we dip into the PC-specific gear, we should familiarize ourselves with the underlying technology. From a hardware geek’s perspective, it’s pretty slick. Thunderbolt is a point-to-point interconnect with an obscene amount of bandwidth and the ability to transmit both DisplayPort and PCI Express signals along a remarkably thin cable. How much bandwidth? 10Gbps in each direction, per channel. Each Thunderbolt connector carries two channels, bringing the pipe’s aggregate bandwidth up to a staggering 40Gbps, or 20Gbps in each direction.

To put that into perspective, a single USB 3.0 port offers 5Gbps of bidirectional bandwidth. One PCI Express 3.0 lane has 10Gbps of full-duplex connectivity. Thunderbolt beats them both with ease. And all of its bandwidth is usable thanks to a low-overhead packet format, Intel says.

Offering gobs of bandwidth is one thing, but Thunderbolt’s real crown jewel is its ability to map DisplayPort and PCI Express signals to a single “meta-protocol.” Translation is handled entirely by the Thunderbolt controller, allowing the interconnect to remain invisible to the host PC, which sees only the connected devices. Software isn’t required for the controller, but Thunderbolt devices will have their own drivers.

Source: Intel

Intel hasn’t revealed exactly how bits are encoded within the Thunderbolt protocol, but we do know there’s a flexible QoS framework tasked with multiplexing the PCIe and DP streams. Depending on the configuration, encapsulated signals may be routed through multiple Thunderbolt controllers before being decoded. As many as seven Thunderbolt controllers can be daisy chained together, and the switch-based architecture supports other topologies, as well.

A domain-based time synchronization mechanism keeps attached devices within eight nanoseconds of each other. That feature will be key to Thunderbolt’s appeal among audio and video content creation professionals. Intel claims Thunderbolt’s low latency as another benefit, although it doesn’t offer any specific numbers on that front.

Source: Intel

The Thunderbolt controller chip does all the work associated with the interconnect. It contains a Thunderbolt switch, a PCI Express switch, and adapter ports for DisplayPort and PCI Express. Controllers can come with one or more Thunderbolt ports, and they connect to the host system using a PCI Express 2.0 x4 link. DisplayPort devices hook into the controller via the associated adapter ports.

Although PCI Express is now in its third generation, four gen-two lanes still offer plenty of bandwidth for a whole range of devices, including the I/O peripherals one might expect to find in a Thunderbolt companion hub for notebooks. A hub could easily offer a collection of Serial ATA, USB, and FireWire ports courtesy of auxiliary controller chips. Heck, you could even throw a discrete graphics processor in there. Indeed, we saw a notebook with a Thunderbolt-connected external GPU at the Consumer Electronics Show earlier this year.

The final component of the Thunderbolt equation is the physical interface: the connector and its associated cabling. Mini DisplayPort has been adopted as the connector, in part because it’s interoperable with existing DisplayPort monitors, but also because it’s small enough for the sort of ultra-slim notebooks that seem ripe for Thunderbolt expansion. Electrical cables can span distances up to 3 m and carry as much as 10W of power for connected devices. Optical cables promise lengths in the “tens of meters,” though the device at the other end will need its own power source. Both types of cables use the same MiniDP connector.

If you’re wondering how Thunderbolt can get away with using an electrical port interface for optical cables, the answer is that all Thunderbolt cables are smart or “active.” They include termination electronics at each end to assist with communications. The optical cables will convert electrical impulses into light at one end and back into electrical signals at the other. The reason not all Thunderbolt cables are optical now? Copper Thunderbolt cables are less expensive.

Despite its considerable potential, Thunderbolt still has some limitations. In addition to the seven-device cap on daisy chaining, no more than two DisplayPort screens can be connected at a given time. Those need to be DisplayPort 1.1a or newer displays, by the way. With older DP screens, users may be limited to one display per chain.

A bigger concern is the lack of hot-plugging support in Windows. If you want a Thunderbolt device to be accessible, it needs to be connected when the system is booted. Intel says hot-plugging support is coming in driver updates for Thunderbolt devices. We’ve heard those drivers may arrive as early as this month and that some cooperation is required from the motherboard firmware. Any Thunderbolt device that’s certified for Windows should support hot plugging, Intel says.

Only premium products need apply

Even when hot-plugging becomes a possibility, Thunderbolt will still be face one very big obstacle: price. The only electrical cable you can buy right now comes from Apple and costs a whopping $50. Imagine what Monster will charge for its inevitable gold-plated version.

Admittedly, there are clear reasons for the high cable costs. As we’ve mentioned, the Thunderbolt cable is an “active” one, which likely raises its manufacturing cost. As a result, non-Apple alternatives may not end up being much cheaper, and the optical variants will probably cost even more. No wonder device makers have been selling their products sans cable. Even the four-drive Promise RAID device we used for testing comes cable-free.

At its Thunderbolt press event the day before the Computex trade show in Taipei, Taiwan, Intel told us that second-generation electrical cables will appear in the latter half of this year. These cables will feature more integration in their active circuitry, and they should cost less as a result, about 25% less, according to Intel. $37.50 for a Thunderbolt cable still sounds pretty expensive to me.

The cost of integrating a Thunderbolt controller chip isn’t trivial, either. MSI told us it costs $30-35 to add Thunderbolt to a motherboard, a figure we’ve heard echoed elsewhere. The controller chip makes up the bulk of cost.

Given the relatively high price of entry, it’s no wonder Thunderbolt controllers have only found their way onto a handful of high-end motherboards. The most exotic of those is Asus’ P8Z77-V Premium, which was the first Thunderbolt-compatible board to arrive at the Benchmarking Sweatshop.

As its name implies, the Premium isn’t cheap. Newegg is selling it for $449 right now, which is more than the cost of a mid-range Z77 board and a fully unlocked Ivy Bridge CPU.

The Premium’s payload certainly lives up to the name, though. This thing is loaded with more accessories than a Harajuku girl. In addition to a Thunderbolt port, it comes with extra USB 3.0 and 6Gbps SATA ports, an mSATA slot populated with a 32GB SSD, and four PCI Express x16 slots. The x16 slots are all connected to the CPU via a PLX switch, ensuring that none encroach on the bandwidth available to the Thunderbolt chip, which enjoys a four-lane PCIe link to the Z77 platform hub.

Asus provided us with some interesting details on how it has deployed the “Cactus Ridge” Thunderbolt controller you’ll find on other PC motherboards. The traces between the Thunderbolt chip and its Mini DisplayPort connector are very short—only two inches long, compared to 10 inches for the board’s USB 3.0 controller. To further improve signal quality, the traces are more widely spaced than usual, and they follow smooth arcs instead of the 45-degree corners used elsewhere on the board.

If the Premium is a little to rich for your tastes, Asus also has the P8Z77-V Pro/Thunderbolt. Unlike the Premium, it hasn’t passed Thunderbolt certification just yet. However, we expect the board to cost around $260.

Folks with select Asus Z77 motherboards will also have the option of buying a Thunderbolt PCIe expansion card. Asus expects the card to cost $40 when it becomes available in late Q2 or early Q3. You won’t be able to plug it into any old motherboard, though. The card needs to see the system’s display output, which it can only do via a separate connection to a special motherboard header.

The next motherboard on our Thunderbolt tour is Intel’s DZ77RE-75K. Like the Asus offerings, it uses Intel’s flagship Ivy Bridge chipset, the Z77 Express. You don’t get nearly as many auxiliary peripherals as on the Premium, of course, but the Intel board is still very well equipped, with integrated Bluetooth, Wi-Fi, and additional 6Gbps SATA and USB 3.0 connectivity. It’s also much more affordable than the Premium, with a suggested retail price of $278.

Note that I didn’t just call a $278 motherboard affordable. It’s more affordable, which is sort of like saying a Lamborghini Gallardo is more affordable than an Aventador.

Intel says the DZ77RE-75K’s Thunderbolt chip gets a full four lanes of PCIe 2.0 bandwidth. As on the Asus board, the controller is extremely close to its associated connector.

While the DZ77RE-75K has completed certification testing, MSI’s Z77A-GD80 has not. MSI expects to have the board certified after the Computex trade show, which runs through the end of this week. If you’re feeling adventurous, you can already buy the GD80 on Newegg. The asking price? $269, or $9 less than the Intel offering.

MSI wasn’t able to get us a GD80 in time for this article, but its Thunderbolt implementation doesn’t appear to differ substantially from the other two boards. They all give the Cactus Ridge chip a PCIe x4 link and a choice location next to the MiniDP port.

The promise of Thunderbolt devices

With two Thunderbolt-equipped motherboards at our disposal, the only missing component is a compatible device. A handful of options are already on the market thanks to Apple’s early entry, and more are on the way now that Windows PCs can get in on the action. Promise supplied us with its Pegasus R4, a four-drive external storage device with dual Thunderbolt ports. The second port allows the Pegasus to sit in the middle of a daisy chain of devices. It can’t get enough power from a standard Thunderbolt cable, though. You’ll need to plug this puppy into a wall socket.

The Pegasus R4 has an understated metal exterior marred by a single piece of glossy black plastic. To be fair, you’re not going to be handling the thing enough to leave behind a mess of smudges. I still deposited a few fingerprints on the plastic panel in the course of unboxing the device and setting it up next to my test rack.

Under the Pegasus’ silver skin sits a PMC Sierra SoC with a Promise “enterprise level” RAID engine. The chip is designed to accelerate RAID 5 and 6 arrays, Promise says, and the Pegasus also supports the usual mix of striped and mirrored configurations. The R4 has four drives, and there’s an R6 variant with six.

Each drive is housed in its own 3.5″ sled. The sleds slide into the Pegasus with ease, and the latching mechanism that holds them in place is satisfyingly chunky. You definitely won’t pop one of these out by accident.

Promise claims the R4 is capable of pushing transfer rates as high as 500MB/s, which falls well short of the bandwidth available in a single Thunderbolt port. Even the Pegasus R6’s purported 900MB/s peak data rate is far from fast enough. Of course, the Pegasus relies on mechanical hard drives; our R4 came loaded with 7,200-RPM Hitachi drives, each with a terabyte of storage. Fill the Pegasus with SSDs, and it should get much faster.

Unfortunately, we don’t have a matched set of four SSDs. But we did spend some time benching the Promise box on the Asus and Intel motherboards. They offer identical performance, as far as our test results show. We considered graphing the numbers, but to be honest, there’s little point. The R4’s performance is limited more by the mechanical hard drives within than by the Thunderbolt cable linking it to the system.

With the R4 in its default RAID 5 configuration, we just about broke 400MB/s with sequential reads and writes in IOMeter. Hitting that speed required hammering the drive with multiple concurrent I/O requests. CrystalDiskMark’s sequential throughput benchmark is limited to a single I/O, and it only reached about 330MB/s. We saw similar transfer rates in FileBench, our scripted Windows file copy benchmark.

Had we switched the R4 into RAID 0 mode, we probably could have hit higher speeds, at the expense of redundancy. We still would have been short of stressing the Thunderbolt link, though. The interconnect is really designed to handle multiple high-bandwidth devices alongside a high-resolution DisplayPort monitor.

There’s another thing, too. We ran the very same tests in our Z77 motherboard round-up using a Samsung 830 Series SSD attached to the Z77’s integrated USB 3.0 controller via a Thermaltake BlacX docking station. In CrystalDiskMark’s sequential read speed test, that combo pulled up just 10MB/s short of the R4 on Thunderbolt. The gap widened in FileBench. However, in CrystalDiskMark’s random I/O tests, the USB-attached solid-state drive was at least an order of magnitude faster than the Pegasus array.

It’s not really fair to compare the two—the SSD weighs in at only 256GB, while the R4 config offers nearly 3TB of redundant storage. But there is an inescapable truth: for most PC users, USB 3.0 has plenty of bandwidth for external storage.

Conclusions

Thunderbolt is one of those things that impresses easily but is difficult to recommend. There’s a lot to like about the technology, which squeezes multiple high-bandwidth protocols into a single cable just four millimeters in diameter. Seamlessly combining PCI Express and DisplayPort is a neat trick. Doing so while delivering 40Gbps of total bandwidth is the sort of thing Steve Jobs might have considered magical.

The real magic is what can be done with Thunderbolt. Take ultraportables, for example. They’re too thin to offer many expansion ports and often lack the chassis capacity and thermal headroom to accommodate lots of storage and powerful discrete GPUs. A single Thunderbolt hub could integrate all those things and provide connectivity for additional displays and more Thunderbolt devices.

In fact, Matrox has just announced a docking station that looks like a good first step in that direction. Unfortunately, you won’t find anything else like it among the Thunderbolt products currently being offered. External storage and specialized audio/video gear dominate the device list, and both appeal to niche markets. While plenty of users certainly need speedy external storage, the vast majority would be better served by USB 3.0 or even eSATA.

If the lack of compelling mainstream devices isn’t enough of an impediment, there’s also the cost. Thunderbolt is expensive to implement on motherboards and in notebooks. The associated cable is pricey, and it’s not bundled with devices. We’re still waiting for optical cables, too. A company called Sumitomo started sampling optical Thunderbolt cables in April, but there’s no word on retail availability or pricing. These optical cables aren’t designed to offer more bandwidth than the existing electrical ones; they’ll just allow for longer cable lengths.

Thunderbolt has plenty of potential for notebooks, where its utility is easy to understand. Since notebooks and mobile devices are a growing portion of the overall computing ecosystem, we expect Thunderbolt will play an important role going forward. Still, it’s difficult to see the appeal  of Thunderbolt for desktop PCs not limited by chassis constraints. Modern desktops have I/O ports up the wazoo, and even integrated graphics solutions can power multiple digital displays simultaneously. The fact that Thunderbolt will likely be included only on high-end motherboards that already feature an abundance of PCI Express, Serial ATA, and USB 3.0 connectivity only makes it feel more redundant.

Comments closed
    • The Jedi
    • 7 years ago

    Wow Geoff, I truly love the blunt honesty here.

    I can see how Thunderbolt would be bigger with notebooks with a discrete GPU and the docking station idea and stuff. It looks like it holds a lot of promise, it’s just that there’s little momentum here thus far.

    • fredsnotdead
    • 7 years ago

    A 4 page quick look! Gotta love TR.

    I’d rather have a slightly thicker laptop with the ports I need (and a larger battery) than the (undoubtedly pricey) non-portable contraption from Matrox.

    • Hrunga Zmuda
    • 7 years ago

    Where did you get the idea that Matrox is the only option. Belkin has one, albeit at $150 more, that not only has what the Matrox Thunderbolt docking station has, but it includes a Thunderbolt out port, three USB 3 ports, eSATA port and a cable pass-through port as well. More money, but vastly more capable.

    And OWC is coming out with some really interesting Thunderbolt devices soon, including a box that lets you plug in PCIe cards. So, a Macbook Air (or a PC imitator with Thunderbolt) can run full-sized PCIe cards when plugged in.

    People who don’t remember how USB was introduced won’t see the parallel here. It took 18 months for USB to become popular. And it was the iMac that finally spurred manufacturers to start making USB peripherals (printers, mice, etc.). For YEARS after the iMac, PC makers were still using ancient ports for mice and keyboards. THIS is why Apple has my loyalty. They don’t drag their feet on new technology. They don’t always get it first, and they don’t always do it best or even at all (Blu-ray), but they do it early more often than not.

      • WaltC
      • 7 years ago

      Apparently, prior to Apple latching onto Intel’s USB standard (that I had in motherboards dating from the late 90’s and the USB drivers that Microsoft supported in a Win98 service pack), Apple had no problem selling its own pre-USB keyboard and mouse technologies…;) But, maybe you can’t remember back that far.

      Unlike with the PC manufacturers, who for years shipped older interconnects alongside USB ports, allowing users to make their own choices, Apple forced iMac owners to go USB …;) Apple removed any choice in the matter. Apple got to USB last, but was certainly first in removing user choice in the matter of the use of USB…;) Lack of choice is, I grant you, an Apple trademark.

      As usual, though, it’s typical to have to remind some people that USB is an Intel-invented standard that was, in fact, used in non-Apple boxes sold before the iMac came to market. Thunderbolt, also, is an Intel invention (not an Apple invention.)

      I don’t own anything with an Apple logo on it, and I’m rather proud of that and have suffered no ill effects at all as far as I can see…;) One of the things I most despise about Apple fanboyism is the “invented here” syndrome–simply because in Apple’s case it isn’t true at all. Yea, Apple was the first to force its users to choose between USB and earlier port technologies, because for Apple it improved the company’s margins by not having to include the older technologies while “forgetting” to pass along the discount to its customers. But, what is that if not typical of Apple universally?

      • End User
      • 7 years ago

      USB 1.1 drove USB popularity.

    • d0g_p00p
    • 7 years ago

    $30 to $35 a chip in bulk, wow.

    • gmskking
    • 7 years ago

    I don’t really see the big point in this. This will go nowhere.

    • Rectal Prolapse
    • 7 years ago

    I hope to live long enough to see a benchmark showing two PCs talking to each other over a Thunderbolt connection. It would be nice to see TCP/IP over Thunderbolt between two PCs – would be great as a replacement for 10 Gbe between two peers – assuming it works!

    With that said, Supermicro now sells Xeon e5 dual-socket motherboards with 2 (two!) 10 Gbe ports for $650 – if only there was a single socket Xeon e3 version of that for a cheap but very high speed file server.

    • dpaus
    • 7 years ago

    Why does that Promise box have an old-school RS-232 port right beside the just-slightly-ahead-of-state-of-the-art Thunderbarf ports?

      • indeego
      • 7 years ago

      Because it’s the most reliable console access to date.

    • jbraslins
    • 7 years ago

    Even if you have thunderbolt capable mobo, you still can’t connect Apple’s Thunderbolt Cinema display to a PC with AMD/NVIDIA PCIe GPU, correct?

      • dpaus
      • 7 years ago

      Then it’s not really a standard, is it?

      • derFunkenstein
      • 7 years ago

      I don’t know why you couldn’t – Thunderbolt needs to see the display output, and if it’s not to carry the display signal then what’s it for?

        • jbraslins
        • 7 years ago

        If the mobo has onboard graphics and thunderbolt, then I’d imagine mobo manufacturer would make it work, but even that is a big question mark. There’s virtually no official comments/information about this anywhere.

        But, my question was about add-on PCIe GPUs. I doubt that GPU will be able to send output signal via thunderbold on mobo instead DVI/miniDP connectors on the back of the card. If it can and will with existing cards (GTX5xx/6xx and Radeon HD xxxx) , i’ll be pleasantly surprised.

        For cards with miniDP on the back (mainly AMD/ATI ones), you can connect prev gen Apple LED Cinema displays that were designed for display port. But you cannot connect Thunderbolt monitors, even tho connector is same shape, making Apple Thunderbolt displays currently usable only with thunderbolt equipped Macs.

      • jbraslins
      • 7 years ago

      This anwers my question. Perfect timing:
      [url<]https://techreport.com/discussions.x/23049[/url<] Pretty simple solution really: [quote<]Thunderbolt needs a DisplayPort connection to function, and each of the boards we've seen relies on the processor's integrated GPU to provide that connection. Users can still run a discrete GPU to a Thunderbolt display by using Lucid's Virtu GPU virtualization software. The TB4 removes the need for Virtu software by integrating a DisplayPort input in the rear port cluster. users can connect their discrete GPU's display output directly to that input.[/quote<]

    • HisDivineOrder
    • 7 years ago

    This technology is a great idea that is in need of some innovative thinking from device/peripheral manufacturers. And a price drop.

    They better hurry though or one of those Wireless spec’s for sending display data wirelessly across a room will sneak up and eat their lunch because the promise of a dock your laptop docks to wirelessly to gain access to bi-directional data transfer of display/kb/m/external storage/possible external video hardware might be just a smidgen more convenient than just faster data.

    I live in hope that Thunderpants is eventually used for boosting the GPU capabilities of increasingly thinner ultrathins/ultrabooks to levels as high as the med-high or highest end video cards, but it’s been a long wait and no one’s going after that market.

    I can’t help thinking that Thunderpants will do much better once they update it to version 2.0. It feels… very beta compared to USB3.

      • asdzxc57
      • 7 years ago

      “They better hurry though or one of those Wireless spec’s for sending display data”

      Exactly. Nobody wants cables anymore. We don’t need another cable. I have everything from RS-232 for PCAnywhere to HDMI. We didn’t need HDMI. I don’t need Thunderbolt. I also Agree with everyone’s comments on how Firewire (which I thought was great) failed.

      Must be tough to be in the cable business. Starting to look like buggy whips.

    • Arclight
    • 7 years ago

    Was ist das?

    • AGerbilWithAFootInTheGrav
    • 7 years ago

    errr Matrox still lives? That is even more magical than Thunderbolt

      • just brew it!
      • 7 years ago

      Just because a company doesn’t make enthusiast hardware any more doesn’t mean they’ve bit the dust.

      Heck, even [url=http://en.wikipedia.org/wiki/Cromemco<]Cromemco[/url<] and [url=http://en.wikipedia.org/wiki/IMSAI<]IMSAI[/url<] (early S-100 PC hardware companies from the '70s) were still in business as of a few years ago (though I believe they have now both folded for good).

    • NeelyCam
    • 7 years ago

    Correction: one PCIe Gen3 lane offers 8Gb/s – not 10Gb/s

    • Thrashdog
    • 7 years ago

    Anybody remember FireWire? Yeah, me neither.

      • Thrashdog
      • 7 years ago

      Wow, the PowerMac-era videographers are out in force tonight. 😉

      • End User
      • 7 years ago

      FireWire 800 smoked USB 2.0. Up until recently FW 800 was my main storage interface on both my OS X and Windows 7 machines (exFAT rocks!). Time moves on and I have now converted over to USB 3.0. with a sprinkling of eSATA.

      I have faith in Thunderbolt. I’ve got a Thunderbolt Display connected to my MacBook Air and I’ve just built a new rig with a Thunderbolt equipped P8Z77-V PREMIUM.

        • superjawes
        • 7 years ago

        FireWire kind of suffered from not having the same sort of support USB had/has. Something about it being a Universal Serial Bus or something…

        Thunderbolt looks nice, especially since it can carry PCI Express and DisplayPort; I’ll give you that. But I think that most consumers are going to stick with what they know works, like HDMI/DVI and USB.

          • Thrashdog
          • 7 years ago

          That was my basic point under the sarcasm; Thunderbolt is technically superior to USB 3.0, but it does more or less the same thing(ie, attach peripherals of various stripes), and USB has lower unit costs and a massive installed base. It seems like Thunderbolt is fighting almost the same sort of battle that FireWire was back in the day.

            • superjawes
            • 7 years ago

            ‘Tis why I meant a little sarcasm with spelling out what USB stands for 😉

            The best bet to get people over USB would be to market it as USB+ or something. Easier to digest, and if you’re really smart, make it USB compatible/compliant. Then you could really market something like Thunderbolt or FireWire.

            • mutarasector
            • 7 years ago

            TB is fighting a similarly uphill battle, but not so much because of a greedy fruity computer company with an exorbitant per-port licensing scheme this time. This time, TB is fighting a fairly well entrenched serial bus that is still widely in use, room for bandwidth upscaling and a humongous base of legacy hardware of which TB simply has zilch.. That said, TB vs USB is a false comparison since TB is a transport layer whereas USB is a serial >bus<. Secondly, TB’s adoption is somewhat dependent on USB itself if it is to survive its legacy deficiency period until such time TB devices become more widely available and ubiquitous.

            Considering TB is also a tad more expensive to implement as it requires direct access to the PCI architecture, and presumably requires PCH access (TB can’t be implemented as a PCI add on card), it appears to me the compelling reasons for it at this point are few until Intel finally gets around to making good on what it originally touted (LIGHT PEAK), whether they introduce ‘Lightening Bolt(tm)” as active optical cables (more likely), or provide electrical<>optical controllers on-die of a chipset (possible, although AMD is likely to be out of the picture).

          • End User
          • 7 years ago

          Let me fix your last sentence:

          “But I think that most consumers are going to stick with the cheap stuff.”

            • Krogoth
            • 7 years ago

            FTFY,

            “Most customers are going to stick with stuff that does the job that doesn’t kill the wallet”

            • End User
            • 7 years ago

            The vocal majority on TR are cheap so and so’s that would freak out if your system cost more that $500. They can’t afford Thunderbolt so they will attack it purely out of spite.

            • wierdo
            • 7 years ago

            You mean practical minded? Yes.

            There’s no point in chasing shiny objects just for the sake of it, a technoloy needs to prove itself more useful for people’s needs than the existing alternatives before it can gain mainstream adoption.

            Until then it’ll be a niche solution resereved for those with exotic system requirements.

            • End User
            • 7 years ago

            I’ve got one cable supplying my 11″ laptop with a 2560×1440 display, FW 800, USB 2.0, audio and gigabit. I’d call that proven technology. I’d call that practical.

            • Krogoth
            • 7 years ago

            That fits the niche bill quite nicely.

        • StuffMaster
        • 7 years ago

        Firewire 400 smoked USB 2.0!

        For me, eSata (and now thunderbolt) is the successor to firewire for external storage.

          • Airmantharp
          • 7 years ago

          Yup. Really, just eSATA for direct storage. If you need more, drop in a card with more ports. Thunderbolt sure looks nice if you need to manage a larger external rack, but it has to fit a niche between a desktop-level use case and an enterprise-level use case that would employ fiber channel.

      • Grigory
      • 7 years ago

      I remember FW by its fanboys. 🙂

      • rrr
      • 7 years ago

      I do. And technologically, it was better than USB, at least pre-3.0 ones. It got killed by an Apple tax.

    • kc77
    • 7 years ago

    It seems like every review I’ve seen has a hard time recommending something that really we don’t need.

    From USB 3.0 to Mini-DP from Mini-SAS to eSATA I struggle to find a use for this technology. I don’t know if it’s really worth it to spend an extra $40 just to have sexy or fewer cables. USB 3.0 is fast enough for any mechanical drive. Even most SSD’s aren’t capable of saturating that.

    I suppose you could use it for ultra thin form factors but it’s not like USB or DP don’t come in a mini format.

      • bcronce
      • 7 years ago

      “Even most SSD’s aren’t capable of saturating that.” yet

      They have already shown that similar SSDs can run 2x-3x faster on PCIe than SATA. TB is just PCIe for the most part. Not only does TB have almost 100% more bandwidth than SATA3.0, but it also has 10x-100x less latency.

      Personally, I don’t think TB is ready for main-stream, with those prices, but a TB *like* device is what we want. I could not care less if TB takes off, but TB is a good idea. I want something like it, preferably an open standard.

        • kc77
        • 7 years ago

        Yes, it did feel like a “640K is more than enough for anyone” type of comment..LOL That being said, SATA is a communication standard for largely HDD and SSD products. TB is not.

        Even if you have a external HDD that communicates via TB, the SATA or SAS standard is still being utilized. The fact that TB has 10 times the bandwidth is irrelevant because until the SATA or SAS standard gets upgraded it’s not enough to saturate USB 3.0. You would need either a top end SSD or maybe some sort of SAS unit where the aggregate read could top the current communication standard. No single mechanical HDD is capable of doing it, SSD’s could.

        However they have a problem…their small size. it’s going to be a awhile until people are going to use SSD’s to backup their storage arrays. The density just isn’t there for that. So this means mechanical…. which USB 3.0 has no problem with. So once again what is this technology for?

        • rrr
        • 7 years ago

        Yep, saturating any available bandwidth with an SSD is deceptively simple, just slap enough NAND chips and a controller capable of handling them. Price of that solution is a different story, but if it’s meant to generate huge throughput, it certainly isn’t meant to be affordable by a long shot.

    • Anarchist
    • 7 years ago

    with thunderbolt is there any reason to continue providing pci-e slots on mobo?

      • pedro
      • 7 years ago

      It’s yet to be seen but I think that those craving extremely low latencies (e.g. those working on many track audio projects) may experience lower latency with adapters plugged straight into the motherboard.

      • Airmantharp
      • 7 years ago

      They provide 8x the bandwidth in x16 3.0 guise?

      • bcronce
      • 7 years ago

      Not once 40/100Gb TB optical ports come out. Intel talked about removing PCIe slots from motherboards in the distant future and instead have a bunch of TB 100Gb optical ports.

      • Usacomp2k3
      • 7 years ago

      Power?

      • just brew it!
      • 7 years ago

      Most desktop PC users aren’t going to want the additional cost (or clutter) of separate external TB enclosures (and the power bricks to go with them) for their PCIe cards.

      • Bensam123
      • 7 years ago

      TB = 20Gbps per channel, max of two channels
      1 Lane of PCI-E 3.0 = 1GBps, roughly 10Gbps
      16 Lanes = 16GBps, roughy 160Gbps

      4.0 will double that again and it is planned to be released in 14/15.

    • Prototyped
    • 7 years ago

    Unfortunately ultraportables these days (aside from the standard “business grade” ones like the ThinkPad X230 — [url<]http://www.engadget.com/2012/05/31/lenovo-thinkpad-x230-review/[/url<] ) are too thin and too "stylish" to incorporate either a proprietary docking station connector (which ThinkPads, EliteBooks and Latitudes still tend to do), or an ExpressCard slot (which at least the ThinkPads offer as options). I do find it interesting that nearly a year and a half after Thunderbolt's introduction, practically the only production peripheral hardware is bulk storage devices. While Intel's implementation of SATA/eSATA doesn't support port multipliers, it would be far less expensive to add a port multiplier capable eSATA port adhering to SATA 6 Gbit/s than to add a Thunderbolt controller. None of these technologies (eSATA, ExpressCard, proprietary docking stations) are as sexy as just a thin cable connecting to DisplayPort, but they are as functional and less expensive to use.

      • teryan2006
      • 7 years ago

      [quote=”Prototyped”<]None of these technologies (eSATA, ExpressCard, proprietary docking stations) are as sexy as just a thin cable connecting to DisplayPort, but they are as functional and less expensive to use.[/quote<] TB has nothing to do with being 'sexy'. A single cable connector is far more reliable (or cheaper to implement reliably) than proprietary dock connector with dozens of pins. Can eSATA carry a DisplayPort signal? No, so it can't be compared. I don't think it's because ultraportables are "too thin and too stylish" to have a proprietary docking station connector. It's because proprietary docking connectors suck in comparison to a standardized connector that can provide the same functionality across multiple notebook manufacturers.

    • Bensam123
    • 7 years ago

    Weird article that almost seems confused. It seems like you want to earnestly recommend TB to everyone, but have absolutely no reason to do so…

    I’m not entirely sure why you would use TB for a docking station over PCI-E either… PCI-E has had hot plugging since gen 1, offers gobs more bandwidth, doesn’t have a pricetag associated with it, and you can pretty much find everything in existence that’s compatible with it.

      • Corrado
      • 7 years ago

      Thunderbolt IS PCI-e. In a cable form for external use, that is.

        • Bensam123
        • 7 years ago

        Then why does it need a bridge chip? It’s not JUST a PCI-E pipe. You wouldn’t need a $85 chip or active cables if that were the case.

        People like to pull the ‘TB is just PCI-E’ argument all the time, but that’s simply not true.

        • mutarasector
        • 7 years ago

        No it is not PCI-e! It’s a >transport layer< that can carry PCI-e, but not a full 16X at this point yet.

          • designerfx
          • 7 years ago

          you forgot “That’s barely implemented, costs extra, and even uses a proprietary cable”. Yeah, way to go intel/apple. way to go. *slow clap*

          $6 USB3 cable or $50 thunderbolt cable with horrible connectors and a painfully restricted cable length?
          what garbage.

      • spuppy
      • 7 years ago

      What I got from it is that it is impressive technology, but not really that useful. Like Bluetooth, it is mostly a solution looking for a problem. That is a vast generalization of course, but overall, neither technology is really worth seeking out, unless you specifically need it.

        • Bensam123
        • 7 years ago

        I wholeheartedly agree. I think what they were trying to do with TB was pretty cool, but how they executed it and brought it to market isn’t going to cut it as there are either solutions already in place (USB) or better solutions available (like the PCI-E docking station).

        I would stipulate that Bluetooth has a niche compared to this tech though. Bluetooth actually brought something to the table, TB isn’t a game changer in any form.

        One has to wonder how much of the $85 chip cost is associated with some sort of blood pact with Apple to keep TB exclusive to their tech.

          • mutarasector
          • 7 years ago

          “One has to wonder how much of the $85 chip cost is associated with some sort of blood pact with Apple to keep TB exclusive to their tech.”

          Unlikely. Intel is first/foremost a chipmaker and is in the business of selling chips. While Apple had a year exclusive head start on everyone, this was never Intel’s game plan for the long run. The exclusivity deal was simply Intel playing the technological class-envy card, and rather sloppily I might add. In one sense, TB has already failed because of this. TB does have something unique going for it because of the Apple exclusivity expiration: It gets a second bite at the apple (no pun intended), or a second chance to make a good first impression It needs to go optical to accomplish that however, since doing so overcomes range/distance limitations, which might be a compelling reason for it.

          • mutarasector
          • 7 years ago

          “I would stipulate that Bluetooth has a niche compared to this tech though. Bluetooth actually brought something to the table, TB isn’t a game changer in any form.”

          Actually, BlueTooth almost died on the vine. Originally intended for PC peripherals, it never really caught on until cell phone makers found a compelling use for it in headsets/earpieces. Bluetooth got a second chance with PC peripherals because of this. Similarly, TB could see the very same stalled adoption, and then an entry into consumer marketplace via a sideways rebirth..

    • Duck
    • 7 years ago

    [quote<]Folks with select Asus Z77 motherboards will also have the option of buying a Thunderbolt PCIe expansion card. Asus expects the card to cost $40 when it becomes available in late Q2 or early Q3. You won't be able to plug it into any old motherboard, though. The card needs to see the system's display output, which it can only do via a separate connection to a special motherboard header.[/quote<] Or if they designed it properly you [i<]could[/i<] plug it into any old motherboard and use it as it without an integrated displayport signal, or provide a displayport connector on there that you can plug into your graphics card's displayport output. There's no reason why the displayport signal has to be routed along the motherboard's PCB.

      • Waco
      • 7 years ago

      This. The implementation is idiotically stupid.

      • HisDivineOrder
      • 7 years ago

      I agree that they should offer the option for “any old motherboard,” but the addition of the DisplayPort and all the circuitry to go with it would drive up the overall cost of the add-on board.

      Seems like they could do both.

        • Waco
        • 7 years ago

        There shouldn’t be a requirement for DisplayPort AT ALL. It’s idiotic.

      • mutarasector
      • 7 years ago

      “Or if they designed it properly you could plug it into any old motherboard and use it as it without an integrated displayport signal, or provide a displayport connector on there that you can plug into your graphics card’s displayport output”

      Can’t be implemented that way. TB requires low level architectural access at the PCH….

        • Waco
        • 7 years ago

        …which it shouldn’t require.

    • dpaus
    • 7 years ago

    Awesome technology earnestly in search of a compelling reason to exist. You mentioned Lamborghinis; it reminds me more of the [url=http://3.bp.blogspot.com/_yvewUf87f8A/TTf_JlKPS6I/AAAAAAAADDg/IkIEmO5ag7I/s1600/xb-70_stbd.jpg<]XB-70[/url<].

      • Scrotos
      • 7 years ago

      I loved the Valkyrie. Awesome plane!

        • End User
        • 7 years ago

        I travelled all the way from Toronto to Dayton at the beginning of May (road trip/new car) just to see the XB-70 at the National Museum of the U.S. Air Force.

          • dpaus
          • 7 years ago

          Wow; I’m planning exactly the same trip soon, except by motorcycle. I’ve been to the museum before, but only left a day in my trip that time, which barely got me through one hanger. I’m planning to spend at least two days this time, with flexibility in my schedule for a third day if I feel like it. How long was the drive?

            • End User
            • 7 years ago

            The drive was under 8 hours from Toronto to Dayton.

            I booked both the Presidential/Experimental and the Restoration tours. Both of those ate up 80% of only day there. 3 days sounds good if you want to take your time and see everything.

      • Krogoth
      • 7 years ago

      XB-70 wasn’t really that exciting. USAF had already done the whole hypersonic, ramjet thing before XB-70. It just another PR stunt of the cold war era.

      ICBMs already made XB-70 obsolete (faster, harder to counter, can be armed with multiple warheads [MIRV]).

        • dpaus
        • 7 years ago

        [quote<]USAF had already done the whole hypersonic, ramjet thing before XB-70[/quote<] Leaving aside the fact that the XB-70 wasn't either hypersonic or a ramjet, in what way had the USAF 'already done' it? The big drawback of ICBMs - then and now - is the difficulty of recalling them.

          • Krogoth
          • 7 years ago

          When things are going Defcon 1, I think recalling is the last thing on minds of generals and commander of chiefs.

            • Airmantharp
            • 7 years ago

            You’re right (says my armchair general experience). B-2 launches are the last stage of ‘negotiable’ offensive actions. If we’re launching ICBMs, it’s over.

        • End User
        • 7 years ago

        Not exciting? It may not have been justifiable but it was one hell of an airplane. I was shocked at how massive it is.

        It reminds of the Avro Arrow (cutting edge/cancelled).

        The only thing that sucked about the museum was that it had a Bomarc on display.

          • Krogoth
          • 7 years ago

          Concord was a spiritual descendant of XB-70 project and Soviets managed to fully developed their own supersonic bomber, Tu-160 which is just about as large as XB-70, but wasn’t as fast though.

            • End User
            • 7 years ago

            Both the Concord and the XB-70 were developed during the same time period (late 50’s, early 60’s).

            Tu-160 was a direct response to the B1-A and not the XB-70. The XB-70 flew 17 years before the Tu-160 had its first flight.

            The XB-70 was designed to cruise at Mach 3. All hail the King.

            • Hrunga Zmuda
            • 7 years ago

            Well, to be accurate, the SR-71 is the king – even today as far as we know. Faster than the XB-70, the crash of one of the XB-70s was in part why it died, but it really wasn’t practical for delivering nuclear weapons. And that’s why the B1 was unnecessary. The only reason we got the B1 was because Reagan used it as a stalking horse to defeat Jimmy Carter in 1980.

            Carter was killing the B1, the reason being stealth technology was coming that would obsolete it quickly. Reagan was hammering Carter for stopping the B1 program, so Carter sent advisors to inform him of the stealth program so that politics wouldn’t get in the way of killing an unnecessary and very expensive project. Reagan didn’t care. He knew Carter would not reveal the reason, so he used that fact to call him weak on defense. So, Reagan’s selfish gain in becoming president was saddling the nation with massive debt because of the expense of a project we did not need.

            Just a little history on why the best technology doesn’t always win.

            • End User
            • 7 years ago

            Well, to be accurate, the SR-71 was not a strategic bomber.

            The XB-70 was not the best tech. ICBMs made that clear.

Pin It on Pinterest

Share This