Thunderbolt Performance

The Eagle Ridge Thunderbolt controller is home to two Thunderbolt channels, each one is good for up to 10Gbps in either direction (up or downstream). That works out to be 20Gbps of bandwidth per channel or 40Gbps aggregate between the two. You can only send two channels worth of data down a single Thunderbolt cable, so there's no point to having more than two from a performance standpoint unless you have more than one port on your system.

If DisplayPort and PCIe traffic are indeed carried on separate channels, then the Thunderbolt Display by itself is eating up around 70% of the bandwidth of a single channel on its own (2560 x 1440 x 32bpp x 60Hz with 8b/10b encoding > 6.75Gbps). That leaves 10Gbps in each direction for PCIe traffic. If we look at our benchmarks from the previous section we see that we can hit just under 2Gbps with all of the auxiliary interfaces (GigE, FW800, USB2) running. Given our previous investigation with the Promise Pegasus we know that 8Gbps is feasible there as well. It's possible, that with everything running at once, we could actually run into bottlenecks with Thunderbolt.

To find out I ran a few tests. First I needed a baseline so I threw four SF-2281 SSDs into the Pegasus R6 chassis and configured them in a RAID-0 array. I ran a 2MB sequential read test (QD=16) and measured 909MB/s from the array. This value was obtained without the Thunderbolt Display connected, only the Pegasus R6.

Next I connected the Thunderbolt Display directly to my test MacBook Pro, and then connected the Pegasus to it. I repeated the test, this time getting 900MB/s. Thankfully the presence of the Thunderbolt Display doesn't seem to impact the max data rate I can get from the Pegasus.

For my third test I added a Gigabit Ethernet transfer from a file server to a local SSD using the GigE port on the display. During this test I was also playing back music using the Thunderbolt Display's internal audio codec and speakers. I re-ran the Pegasus test and got 855MB/s.

For my final test I re-ran the third test but added a FireWire 800 to USB 2.0 SSD transfer, both connected to the Thunderbolt Display. I also fired up the FaceTime HD camera on the display using Photo Booth and left it on during the test. The final performance score from the Pegasus was 817MB/s.

Apple Thunderbolt Display Performance

With everything running Thunderbolt performance took a 10% hit. Note that the standard Pegasus configuration isn't able to hit these data rates to begin with, so unless you've pulled out the 12TB of storage and stuck in your own SSDs you won't see any performance drop.

What this does tell me however is the ultra high end users that are looking to daisy chain multiple Thunderbolt storage boxes together may not want to do so. I only have a single Pegasus R6 on hand, but I'm guessing there will be significant performance drop off after the first box. Not that I'm complaining about being able to push nearly 1GB/s over a $49 cable from a notebook, I'm just trying to give a heads up to those who may have aspirations of even higher performance.

Testing the Pieces Display Testing - Color Quality & Uniformity
Comments Locked

275 Comments

View All Comments

  • Constructor - Tuesday, September 27, 2011 - link

    You're comparing USB to a GPU? ok...


    No. The former is an interface protocol, the latter a component.

    But Thunderbolt can work on a completely different level with GPUs than USB can:

    Internal GPUs are usually connected to the CPU through 16 PCIe Lanes, which adds up to 64Gb/s in each direction with PCIe 2. These GPUs also have the necessary power and cooling available for advanced 3D performance.

    And at their outputs, they're pumping out substantially more than 5Gb/s for a high-resolution video signal.

    USB3 cannot supply an external GPU with enough data throughput to come even close to an internal GPU. The outright horrible inefficiency of the USB protocol may well eat up half of the nominal bit rate again as it has with earlier incarnations. It also has too little bandwidth to tunnel the video output signal from an internal GPU, let alone on top of full-speed data transfers at the same time.

    Thunderbolt on the one hand provides at least double the data bandwidth of USB to drive an external GPU if so desired (which might actually be enough for mid-range 3D applications) and with Thunderbolt that is actually the net throughput with next to no overhead, and on the other it piggybacks not just one but even two complete high-resolution video streams onto the same cable at the same time if so desired!

    The two are playing in completely different leagues – USB falls far short of what Thunderbolt can do, not just theoretically but in actual practice.

    You will never see anything like the Thunderbolt Display with USB3, at least not remotely with the same performance since it is simply not possible.

    I have no perceivable latency on my USB mouse, and I'm sensitive to it.


    Sorry, but I think you don't understand what I am talking about. That you may not be able to perceive the latency of your USB mouse with the naked eye is as it should be, but extremely far removed from the micro- or even nanosecond-range latency requirements of many more advanced interfaces (human perception begins only to notice differences starting in the millisecond range, usually even just at tens or even hundreds of milliseconds).

    There are good reasons for advanced audio and video interfaces insisting on using FireWire so far and now starting to switch to Thunderbolt. USB has always had extremely bad response latencies which is inherent in its protocol specification. Critics even suspected that Intel had screwed up the USB protocol that badly because they wanted to keep their CPUs busy so they could keep selling faster ones, even though that is probably just a myth... 8-)

    Theoretically USB3 could improve on that to some degree (which is a completely different issue from sheer data throughput), but we'll have to wait and see whether that will actually come about.

    "USB is a...not too fast peripheral interface." - 5Gbps isn't fast? No external connection aside from raw video can utilize that![/b]

    USB has so far wasted about than half of its nominal bandwidth idling in its extremely inefficient and wasteful protocol. When have you ever had true 480Mb/s (60MB/s) transfer rates on USB2? Right: Never.

    USB3 may again improve over the older variants, but it's to be seen how much of these problems will actually get solved in practice.

    With Thunderbolt the 10Gb/s are effective throughput, right out of the gate. And that's even just half the total throughput, with the other half tunneling Display Port.

    Thunderbolt plays in a completely different league from USB, even USB3.

    Face it, you don't know what you're talking about.


    Maybe. Maybe not... ;-)
  • Constructor - Tuesday, September 27, 2011 - link

    Sorry for the screwed-up boldfacing above. Post editing would be really handy there...! B-)
  • AnnonymousCoward - Tuesday, September 27, 2011 - link

    You're mixing the topics of GPU processing and the interface to the monitor.

    Some facts:
    -There's no question that USB3 is not suitable for driving high resolution raw video. Throughput maxes out around 3.2Gbps. 1080p60 24-bit requires about 3.6Gbps.
    -Thunderbolt just combines the display and data connectors. When comparing this to a USB solution, compare it with USB3+DisplayPort.
    -No external connection can saturate 3.2Gbps. No hard drive, no web cam, no sound card, no peripheral. So Thunderbolt provides no benefit in its 10Gbps vs 3.2Gbps ability.
    -Thunderbolt gives you few ports and forces you to daisy-chain. And some Thunderbolt devices only have 1 port, meaning they stop the chain. Having 4 USB3 ports in 1 place is more centralized/convenient.
    -Intel probably charges $15/chip. Versus $2/chip for USB3 hosts. The cables are also far more expensive. And for what benefit? To combine data and video cables and force you to daisy-chain.
  • repoman27 - Tuesday, September 27, 2011 - link

    Some corrections to your "facts":

    -Plenty of external devices can saturate USB 3.0. For instance: single current generation SSD's, SSD or HDD RAIDs, 10 Gigabit Ethernet adapters, Fibre Channel adapters, pro video gear, any number of medical or research lab devices.

    -No Gigabit Ethernet adapters exist for USB 3.0 yet, so you would be limited to < 320 Mb Ethernet connections for a while.

    -No USB to FireWire adapters exist. USB does not support peer to peer transfers the way FireWire or Thunderbolt do, and thus performance drops off dramatically when copying between two devices connected to the same root hub.

    -Some USB devices also block a port... actually most do.

    -Thunderbolt doesn't actually force you to daisy chain. Although the silicon doesn't exist for them yet, in theory Multi-port Thunderbolt switches are quite possible.
  • AnnonymousCoward - Tuesday, September 27, 2011 - link

    Your points don't apply to 99% of users.
    -500MB/s SSDs are internal drives. All external drives I'm aware of are platter-based. RAID-away with platters, and you'll still have a hard time hitting 400MB/s.
    -10Gb ethernet? Most people don't care about LAN traffic, and hard drives are often the bottleneck anyway at 90MB/s.
    -Fiber equipment and pro video gear--1% of users?

    You said it doesn't force you to daisy chain; when you buy a CrapBook Pro, you get 1 port. That means you have to daisy chain.
  • repoman27 - Wednesday, September 28, 2011 - link

    "Facts" aren't facts if they're wrong even 1% of the time though. While all of the examples I cited may be corner cases, they are all valid and contradict points which you presented earlier as factual statements.

    Also note that among the 1% of users that can benefit greatly from technologies like Thunderbolt are quite a few of the folks that create the music you listen to, produce the movies and TV shows you watch, make the products you purchase and use every day, develop treatments for illness, and help us to better understand the world we live in. Even if you resist adoption of new technologies personally, you can still benefit greatly from them.

    Apple has long had the philosophy that if you put advanced technologies into consumer devices that the users will find creative and unforeseen ways to exploit these new capabilities. I don't think they're going to give up on this just because you happen to disapprove.
  • AnnonymousCoward - Wednesday, September 28, 2011 - link

    Ok, you are technically right that "No external connection can saturate 3.2Gbps" isn't an accurate blanket statement. It was intended to be effectively true for almost all users. Sure you can make a $5000 RAID array that reaches that. Many people seem to think Thunderbolt offers a speed advantage over USB3 in applications where it doesn't (nearly all applications).

    You may be overstating Thunderbolt's benefit for that 1%. So it lets you dock a laptop without a normal docking station, or it saves you from plugging in wired ethernet (yet it doesn't save you from plugging in wired analog audio). That isn't going to help any artist do their work. The 1 unique capability I see that it offers, is the bandwidth to download raw video to a PC. But PCIe cards probably already exist that receive HDMI or DP for a computer.

    And I have a feeling that if James Cameron had a Thunderbolt-enabled computer and associated equipment, Avatar would have sucked just as badly.
  • repoman27 - Thursday, September 29, 2011 - link

    Believe it or not, there are actually people who can use (and are excited about) the little known already shipping Thunderbolt devices such as: http://store.apple.com/us/product/H7293LL/A/promis...

    And I'd be willing to venture that your feeling about Avatar constitutes an actual fact.
  • Mystermask - Sunday, October 9, 2011 - link

    Unbelievable how stupid so called "tech enthusiasts" behave when they read the Name "Apple".

    1. We've been reading the same technophobic "you don't need that" comments in the past about almost any switch Apple made: the replacement of proprietary ports with USB, 5 1/4" floppys with 3", 3" floppys with CD, CDs with DVDs, cell phones with physical keyboards with Multi Touch devices, etc
    2. Let the market decide what is needed. E. g. I'm always astonished how software / solutions make creative use of given hardware capabilities. This kind of evolution can never happen when HW vendors only stick to "what 99% of users need today".
    3. By your standards, we should stick to old HW (CPUs, GPUs, ATA, PS2, ..) because 99% of all users don't need newer stuff: They surf, write e-mails, watch photos or use bloated Word from time to time. No
  • Jamezrp - Monday, September 26, 2011 - link

    The video's really well done. I'd recommend using some of the YouTube features for links at the end, though it's clear you did this yourself even though the production value is clearly high up there (good camera, good audio, but minimal editing and control). Great stuff.

    That said, I've been pondering the Bluemic for some time and the quality on this video just sold me. Completely wrong product...I'd love the Thunderbolt Display to go along with my 13" MBA, but I've already got a Dell 2408 and Acer G24 for my PC which takes up my entire deskspace, and don't need the MBA on a bigger display anywhere else.

Log in

Don't have an account? Sign up now