USB 3.1 vs Thunderbolt - the battle is on!

The USB 3.0 Promoter Group on Wednesday announced that it has completed the USB 3.1 specification. Most notably, USB 3.1 supports data transfer speeds of up to 10 gigabits per second (Gbps), the same raw transfer speed as the Thunderbolt interface found on all currently shipping Macs. The group is sponsoring developer conferences in the U.S., Asia and Europe to spread information about USB 3.1, according to a press release.

USB 3.1 will supplant "SuperSpeed" USB 3.0, which is limited to 5 Gbps. The USB 3.0 Promoter Group says that more efficient data encoding is used in 3.1, enabling the higher speed. As before, USB 3.1 will be backwards-compatible with older USB devices.

USB 3.1 will see widespread adoption as a common computer interface, just as other implementations of USB have. And the "USB 3.1 matches Thunderbolt speeds" trope is in full swing at this hour on tech blogs around the globe. But it'd be an enormous mistake to discount Thunderbolt, notwithstanding Thunderbolt 2's imminent arrival.

Thunderbolt cable

Think of USB 3.1 and Thunderbolt like roadways. Both have the same speed limit: in this case, 10 gigabits per second. But USB has only two lanes on which traffic can travel, one lane of traffic in either direction. Thunderbolt has four: two up, two down - each of those with 10 Gbps of effective bandwidth. What's more, Thunderbolt has considerably better traffic management than USB does. The net result is that Thunderbolt is still going to win out in real-world usage.

There can be only one...or not

I have absolutely no doubt that USB 3.1 will ascend as the heir apparent. USB 3.0 is already ubiquitous; Thunderbolt, meanwhile, has had a very slow adoption rate outside of the Macintosh platform. I don't expect to wake up tomorrow and find every shipping PC with Thunderbolt ports. It ain't gonna happen.

Ultimately Thunderbolt and USB 3.1 are really different interfaces aimed at different market needs. Thunderbolt provides a small, high-speed interface that can easily replace other interfaces and works fast enough to push video without any problem (and once Thunderbolt 2 arrives, we'll be able to do 4K video).

USB 3.1 provides further improvements to a ubiquitous peripheral interface that supports a huge number of legacy peripherals - an interface that's actually mandated as necessary for some peripherals like cell phones, by various governments and regulatory bodies around the world.

USB 3.1 isn't going away.

It'll take a while to bake

Then there's the issue of availability.

Now that final details on the spec are available, vendors can get to work. But don't expect USB 3.1 hardware to appear right away. The USB 3.0 Promoter Group doesn't expect to see USB 3.1 devices appear even with limited availability until late in 2014. Between now and then, CPU makers like Intel and AMD will need to incorporate support for the faster spec into their chip designs. Then those designs will need to go into production. So it'll be a while before we see the market flooded with USB 3.1-compliant devices.

In the meantime, Apple is working on a radically redesigned Mac Pro, which it expects to ship later this year. Among the many innovations is Thunderbolt 2, an updated version of the high speed peripheral interface found on Macs that delivers twice the bandwidth - up to 20 Gbps. By the end of 2014, expect to see Thunderbolt 2 widely available on the Mac platform.

Is USB 3.1 a big deal for you? Did you notice a big difference when you used USB 3.0 for the first time, compared to 2.0? Or are you loyal to Thunderbolt? Share your thoughts below.

Peter Cohen
  • I think for many it's going to be a no-brainer. USB 3.0 or 3.1 For example, I need to add a basic RAID to my mac mini (2011). As I've toyed with various options, it actually ends up being overall more cost effective to just buy a new mini with USB 3 ports and add a couple of 3TB drives I can buy off the shelf at any store for ~$150. (and sell the 'old' mini). I need more speed than USB 2, but I don't need TB speed. The TB stuff is just WAY too expensive yet to really compete. Now, for the higher end, yes, I suppose it starts to make sense. If I were doing an SSD RAID or a bigger 5+-disk RAID, I'd probably lean towards TB, but would also have to have the budget to go along with that. The problem with TB, on the other side, is that it just isn't fast enough yet. Even TB2 won't be fast enough to really replace an internal bus connection. So, it's a 'PRO' connection that really isn't up to the task outside of storage or some peripherals, of which many of the latter would also work via USB 3. So, it's better... it just might not be enough better to matter. Once Intel gets it to the final stage, it might be.... but who knows what the competition will be by then.
  • I disagree on a few of your comments. "The problem with TB, on the other side, is that it just isn't fast enough yet. Even TB2 won't be fast enough to really replace an internal bus connection.". Im not sure where you have the idea TB has to be as fast as an internal bus to be "fast enough". For me that bandwidth trumps having to be as fast as an internal bus. We have development teams and our TB monitors power everything. Besides most Macs come with both. On Windows, USBx is more common.
  • His point is the following: 1) For the vast majority of users, TB2 is "fast enough." 2) For many of those same users, though, the USB3.x is also "fast enough" -- and cheaper 3) The users who do not care about cost differential also do not care about "fast enough" -- they want "as fast as possible" to do their work. (Don't think monitors, think RAID arrays in a video studio -- there, you need as fast as you can get.)
  • Pretty much... :)
    Only for #3, I'd say more for things like GPU addition to the Pro machines, like the upcoming Mac Pro. TB2 should be fast enough for RAID, but it won't be fast enough to add really fast cards or GPUs externally. But, yea Ivavila, my statement was probably a bit overblown, as there certainly is some mid-range gray area in there.
  • There are very few peripherals for which TB2 will be unsuitable. The big one would be external GPUs. This is why the MacPro has 2 standard internal. Correct me if I'm wrong, but with the current OS, Mac's can't even recognize more than 2 GPUs. The current RedRocket card can run off Thunderbolt, but the second generation RedRocketX will probably be too much for it. Basically anything that's not a 16x lane PCIe card should be able to run in an external chaises. And the rumour is that you can multiplex TB2 ports, so with 3 TB2 controllers in the new MacPro, the useable bandwidth could be even greater.
  • I should probably work the math through, but I think it's more like anything over 4x lane PCIe. Combining TB2 ports would be interesting if that's possible. I'm still trying to figure it out, but the problem might be more the ATI than how many, if CUDA is an issue for certain Pro apps. Maybe Apple's got OpenCL all ready to go. We'll have to see, I guess. Of course, then there is the cost, but that's a whole other discussion. My main point was that for most people, USB 3x is going to be fast enough, pushing TB2 towards the more 'Pro only' crowd... for which it's a bit iffy if it really is going to be fast enough for everything they need, in order to become popular enough for survival. If it really were here in it's full optical goodness (like 2015 or so?), then I'd say it is such a night and day difference and would fill every pro wish or need, that it will certainly succeed. But, it's less certain now, it seems to me.
  • I've read a couple of conflicting comparisons trying to measure TB2 against PCIe (v3.0). Here's one that looks fair- A single-lane PCIe (v3.0) card is capable of 985 MB/s (bytes) in each direction. That’s 7.88Gb/s (bits). An 8-lane PCIe card is capable of 7.88*8=63Gb/s — more than 3 times faster than TB2. Put another way, TB2 is akin to a 2.5 lane PCIe card (if there was such a thing), NOT an 8-lane card. There are 6 channels @ 20gbps on a new Mac Pro… combined throughput akin to a SINGLE 16-lane v3.0 PCIe card (126gb/s). --------------------- So no question that TB has a generation or so to go before being the internal bus replacement it needs to be to handle ALL cards. But as I've said elsewhere, where's the need for 16x cards when the MacPro comes with 2 GPUs? Blackmagic, Adobe, and Autodesk have all said that their relative apps are OpenCL compatible now, with little to no performance difference to a comparable CUDA card. They've all said it's a monster. If multiplexing TB is true, then there's a very narrow window of PCIe peripherals that won't run.
  • Thanks for that comparison... sounds about right to me too.
    But, from what I've heard *today* must mean in their test labs... so hopefully that's accurate. Yes, I guess if CUDA becomes irrelevant, then there isn't much need for 8x and 16x slots... so most other stuff could be connected via TB2 (at a cost, of course).
  • Wrong headline, USB vs Thunderbolt battle is already over, USB won, thunderbolt will always be a niche interface whereas USB is the defacto standard.
  • The headline got you to read the article and comment on it, so I think it's just right. ;)
  • Thunderbolt is too expensive.
  • No. You just can't afford it. There are cars you probably can't afford either, are they too expensive?
  • This is a bit different though... the analogy kind of falls apart. If TB doesn't become popular enough, it's eventually not going to make it. If some Bugatti doesn't become popular enough, that makes it even more sought after.
  • What's successful enough? Relative to USB2, Firewire NEVER gained widespread appeal- relegated mostly to prosumer and pro markets. As 4K becomes more the standard for all areas of video production, not only being able to play that material back, but also just move it around... I think people will want the 2X+ performance that TB2 provides. Based on what Apple has done with the MacPro (throwing all expansion outbound), that says to me that Apple believes TB will get to it's 100Gbps spec in relatively short order.
  • If my car cost significantly less but does practically all the same things - even if it doesn't look quite as slick or have as marketable a name -then yes it is too expensive and people won't buy it. I can afford to pay twice as much, but what would be the point?
  • If that was true everyone would be buying the cheapest car they get- which almost never happens. There are always modifying factors- one person has 3 kids, another puts value on better fuel economy, and one wants a pickup even though he doesn't really need it. And the truth is that USB3.1 doesn't do all the same things. It can't drive higher end displays, it's not as good for media because it's not a sustained bit-rate, AND it's half as fast.
  • To answer the question posed, yes, I notice the difference in USB 3.0 over 2. But mainly only when I am booting into OS X installed on a USB 3.0 enabled drive attached to my MBA, and hardly can tell I am not on a native drive. Sent from the iMore App
  • Obviously the best isn't good enough. It seems "good enough" is what usually wins. People use cell phone cameras instead of pocket cameras, and pocket cameras instead of DSLRs. "Good enough" usually wins cause it is cheaper. And with USB's ubiquity that just means that if you want the best then you will pay a premium because the TB devices won't sell enough units to create price wars. I love the speed of my TB drives. I would love to double that speed. But the price for TB is going to be very hard to justify if USB 3.1 meets the hype. Is that 3.1 rate burst or sustained?
  • Burst rate, much like existing USB3 specs. So continuous data transfer won't be anywhere near as good as TB1. But I don't get the Highlanderism here [thanks AMB!], Macs ship with both. When Intel puts out a chip that supports USB3.1, Apple will adopt it. Meanwhile if you need 2x the bandwidth, TB2 will be there for you. Don't buy it if you don't need it, but for pete's sake stop complaining about how much the bleeding edge costs. Much of the comments I've been reading on this today just smack of so much entitlement.
  • It's not entitlement... just a bit of sadness over bad planning and marketing. TB, like FW, should have killed USB. It certainly is superior as an interface. But, if it remains a 'pro only' kind of interface, all the TB stuff is going to stay quite expensive until it eventually is replaced by something else or goes away. How does this affect me? Well, first the cost. But second, I still have FireWire devices around, for which I have to get adapters because there are no FireWire ports on Macs now. My USB stuff just plugs in. I'd just hate to see the same thing happen to TB in a few years.... and I kind of hope to actually see the full optical bus-speed version one day. (ie: I'd like TB to survive.)
  • The "USB will kill Thunderbolt" or "Thunderbolt should have killed USB" arguments are, to borrow a phrase from my buddies over at Angry Mac Bastards, "Highlanderism" (referencing the 80s action movie). There doesn't need to be only one. Thunderbolt and USB can live in harmony.
  • Not only does there not need to be only one ... but it is better when there isn't. USB had stagnated. TB kicked it in the pants and now we have competition, innovation, and price pressure. All good for the consumer, yes?
  • I think USB already won. No matter what speed thunderbolt has, everyone uses USB. Only few people, who wants the speed of the thunderbolt uses it but otherwise, thunderbolt is done. Do you find thunderbolt in PCs? No, I've only seen them in mac which holds about 5% of the marketshare in desktop area. The point of data sharing, means it has to be compatible with other devices. It's just like PC and mac. Microsoft word is not compatible with pages. Samething with USB and thunderbolt.
  • If marketshare was all that mattered Apple would have folded up a long time ago. Specialized faster I/O is very important to a certain sector of the computer market. TB doesn't have to be the popular choice to be successful. It just has to serve it's intended market well. It certainly does for me.
  • My money is on more wide adoption of TB. Now if you'll excuse me, I have a laserdisk to watch.
  • Why is there a Thunderbolt to begin with? Didn't Apple learn its lesson with FireWire? The whole market is USB, connectors are cheaper to produce, its backwards compatible, there are USB connectors and devices everywhere, do we really need another port standard? Is there so much available room on tablets, phones, notebooks and desktops so we have to worry about two ports that do basically the same thing? Sure, there is speed, but how did that work in real life for Firewire, did most people achieved the speeds Apple promised? And what about daisy chaining? So all devices MUST have two ports or we're screwed? What if you want to disconnect the device at the beginning of the chain? Or what if the first device is a Thunderbolt 1, how's the speed on the other Thunderbolt 2 devices down the chain? On USB you don't have to worry about any of that.
  • I've never even experienced USB 3.0 or Thunderbolt. I rarely ever even touch a computer anyway since moving on to tablets and smart phones.