Comments Locked

25 Comments

Back to Article

  • werfu - Wednesday, January 4, 2012 - link

    As thunderbolt is a PCI-e passthrough and that PCI-E is hotplug, I'm pretty sure we'll see cheap adapter come to the market soon.
  • zorxd - Wednesday, January 4, 2012 - link

    Still, if it's not integrated in laptops it won't succeed, just like Firewire.

    It's cheaper to use eSATA for hard drives and HDMI/DVI/DisplayPort for displays. For just about anything else, USB2 is good enough.

    The only advantage of Thunderbolt is to make displays that include many features such as ethernet and USB with only one cable. Will it be enough? I don't think so, personally.
  • solipsism - Wednesday, January 4, 2012 - link

    FireWire's downfall was the high cost of entry. It also didn't have Intel backing it the way Intel backed USB. Don't think that because TB was exclusive to Apple for a year that it's an Apple tech with high tariffs.

    No, I don't want eSATA to connect an external SSD/HDD. I want something that supplies power.

    I also like the fact the port for an external display has a secondary purpose for protocol agnostic data because the majority of notebook users do not use the video out for the life of their machine. Still, the USB-IF missed their chance to get TB integrated with the USB port interface which would have been better for them in the long run.
  • SleepyFE - Wednesday, January 4, 2012 - link

    Some eSATA ports are a USB combo and therefore do have power.
    Also TB can reach higher speeds but unless you have you SSD in a PCIe it doesn't matter. SATA III can go up to 6Gb/s so it makes it a bottleneck. USB 3 on the other hand can make 5Gb/p which is close enough and you can plug a USB 2 device in it (and we all have those).
    My point is that USB is catching up to SATA regarding speed and that is enough to use it instead of TB. The only thing i don't like is the fact that boards still have both USB 2 and USB 3. FFS!!!! USB 3 is backwards compatible. Is it really cheaper to use a USB 2 controller PLUS a USB 3 controller instead of just USB 3?
  • zorxd - Wednesday, January 4, 2012 - link

    All hard drives are SATA internally anyway. There are no TB (or USB, or Firewire) drives.
  • SleepyFE - Thursday, January 5, 2012 - link

    Not all drives. You can get an SSD with PCIe plug, OCZ even made a hybrid drive with PCIe.
  • Kristian Vättö - Thursday, January 5, 2012 - link

    There is always SAS and IDE but in terms of today's mainstream HDs, you are right.

    Although, I remember that there were a few HDs with internal USB. They were used in external HDs and mainly to save the price of SATA>USB bridge. I know I'm just nitpicking :-)
  • zorxd - Wednesday, January 4, 2012 - link

    you can have eSATAp, which is powered.
  • ltcommanderdata - Wednesday, January 4, 2012 - link

    eSATAp is not an official standard and is not currently supported by either the USB or the SATA working groups, which doesn't make it an ideal competitor to Thunderbolt.
  • ltcommanderdata - Wednesday, January 4, 2012 - link

    eSATA doesn't supply power. In a portable scenario it's much more convenient to connect a HDD using a single cable for both power and data. As more laptops switch to capacity limited SSDs I can see more people carrying around a portable HDD while on the go. I believe 3.5" HDDs often require more power than USB 3.0 can supply on a single plug, so Thunderbolt does has some advantages there.
  • name99 - Wednesday, January 4, 2012 - link

    Negatives for eSATA:
    - no power
    - hubs are (as far as I can tell) possible but rare and expensive
    - how common is is REALLY? I don't track PC laptops in detail, but my impression is the few of them have eSATA. For example, I think I'm correct in saying that none of the announced ultrabooks have eSATA.

    Positives for eSATA:
    - (Maybe) What is the expected story for how TB connects to storage? Is the TB conduit acting as a SATA conduit, a SCSI conduit, or a USB conduit? The reason I ask is: will NCQ, TRIM, SMART and whatever other new command-sets area added to ATA be passed through all the way to an external drive, or will we see the sort of crap we saw with USB, where a (largely useless) "abstraction layer" sits between the computer and a TB HD?
  • CharonPDX - Wednesday, January 4, 2012 - link

    TB connects to storage over PCIe. Period. TB is just PCIe in a cable.

    Storage devices must have their own PCIe storage controller chips. For example, the LaCie Big Disk Thunderbolt uses a plain vanilla SATA adapter, no RAID. So the two internal drives in the chassis appear to the host OS as two separate drives connected to a SATA controller, that SATA controller connected via PCIe.

    The Promise Pegasus RAID Thunderbolt, on the other hand, has a full-on SATA RAID controller. So you have to use RAID configuration software that configures the PCIe RAID chip to recognize the drives however you want.
  • MySchizoBuddy - Wednesday, January 4, 2012 - link

    you are forgetting that one potentially NEW market is for external GPUs, just like external harddrives. TB is perfect for it. Imagine 2-3 GPUs being sold as compute nodes to existing systems via TB. notebooks at home can be turned in compute monsters while remaining light on the road.

    I for one am very much interested in such a product
  • r3loaded - Wednesday, January 4, 2012 - link

    It's certainly a possibility, but 10Gbit each way will bottleneck nearly all graphics cards.
  • CharonPDX - Wednesday, January 4, 2012 - link

    Yes... Because first-generation PCIe SLI motherboards were severely bottlenecked...

    Thunderbolt supports up to four channels of PCIe 2.0. It is generally configured as two channels each direction. These CAN be linked together to provide the equivalent speed of a PCIe 1.1 x4 slot.

    Yes, the total bandwidth is less than an internal slot - but for a compute node, it's not likely to affect performance by so much that it becomes not worth it. Even for external graphics use, a GeForce 580 or Radeon 7970 would still be significantly better than ANY internal notebook GPU.
  • Migelo - Wednesday, January 4, 2012 - link

    TB is 20Gbps in each way, making it equal to x5 PCIe 2.0 slot. (20Gbps/8=2,5GBps; 2,5GBps/500MBps=5)

    But how much does a decent GPU really need in terms of bandwidth, like HD69xx or GTX570/580??

    Because it would really be cool of you could buy an external GTX570 and connected it via TB to a notebook/ultrabook and a dedicated power cable to an outlet and with some driver magic :) get a monster ultrabook. Because notebooks/ultrabooks are more GPU than CPU bound if you ask me.

    What do you think?
  • Kristian Vättö - Thursday, January 5, 2012 - link

    I don't think the market for external GPUs would be that big, though. Most people are fine with the integrated Intel graphics. eGPU enclosure would cost hundreds of dollars which would definitely drive it away from mainstream market. I can only think of a gamer prosumer who wants an Ultrabook to be his main computer. In that case, an eGPU might be useful but almost in all other cases, I think it would be cheaper to just build a PC and buy a cheaper laptop.
  • piroroadkill - Thursday, January 5, 2012 - link

    Not quite true, since it includes video output generated by a graphics card - DisplayPort.

    Other than that, your point is solid enough. I'm guessing cable length concerns have made Intel develop thunderbolt controllers as opposed to just having a PCIe lane in the connector directly.

    It would be cheap and simple to implement this way, and require no extra silicon..
  • cyabud - Wednesday, January 4, 2012 - link

    Thunderbolt peripherals can also be daisy-chained to a single tiny, speedy port. Good news for ultrabooks, no doubt.
  • Exodite - Wednesday, January 4, 2012 - link

    Point being, what Thunderbolt peripherals?

    All this is frankly unsurprising.

    Given that we already have USB, which is both forwards- and backwards-compatible and enjoys a frankly massive support, it's not unlikely that the standard will have improved to the point of making Thunderbolt redundant by the time the latter have built up enough momentum to matter.

    Thunderbolt should have been USB 4.0, using the same connector and offering full backwards compatibility with previous USB versions.

    Piggybacking on mini-DP, which is already a fringe standard in it's own right, doesn't help.
  • Shadowmaster625 - Wednesday, January 4, 2012 - link

    Is it thunderbolt or thunderbold? lol.

    I am waiting on some sort of reasonable single wire docking solution for a complete workstation with dual/ triple monitors. I have 5 figure purchase order waiting for the right solution. Sadly nothing delivers, not for a resonable price anyway. I define reasonable as:

    i5-2410M, 256GB SSD, and of course the docking solution (two of them actually) with the single wire interface.

    Is it really too much to ask to get that for $1000? I dont think so since I can buy the notebook for $550, and the SSD for $300, leaving $150 for a pair of docking solutions.
  • phatboye - Wednesday, January 4, 2012 - link

    As long as Thunderbolt remains a proprietary standard I will not be investing any of my money into it. Rumors have it that the PCI-sig plans on working on a extermal-PCI that will compete against Thunderbolt for this reason. If not USB 3 will hold me over for the next few years until something else comes along.
  • jontech - Wednesday, January 4, 2012 - link

    That thing is a beast

    With my EPP I got for $1775 from Apple and there is nothing that comes close to it for that size and speed.

    Thunderbolt takes the need for laptops to have so much internal memory and places the onus on the external

    A shot in the arm for those of us who want speed and storage.

    USB 3.0 Cannot compete at this level, though its great for archive purposes.
  • chizow - Thursday, January 5, 2012 - link

    Is it means Intel can finally stop dragging its feet with native USB 3.0 support. Its obvious Intel wanted to give their proprietary Thunderbolt a fighting chance by keeping USB 3.0 at arms length and off their chipsets, but thankfully, OEM board makers pushed adoption by using controllers from NEC and others to force Intel's hand.

    I hate what Intel is doing on the chipset/socket side of thing though, ever since P55 we have a long list of very incremental and minor updates just so Intel and their partners can sell new boards every year.
  • danjw - Thursday, January 5, 2012 - link

    They can't really believe this interface will be successful at this price point? I just don't see how the development they put into this interface is ever going to pay off for them. Though, I believe they intend to use it at some point to communicate between chips. If this has any chance of being widely adopted, they need to work on getting the price down.

Log in

Don't have an account? Sign up now