Home / News / Peripheral News / Intel Talks Thunderbolt - Supports USB3

Intel Talks Thunderbolt - Supports USB3

David Gilbert

by

Intel Talks Thunderbolt - Supports USB3

With promises of searing transfer speeds of up to 100 gigabits per second – in both directions – Intel’s Light Peak (or Thunderbolt as it is now called) was always going to attract attention, and the computing giant has gone into some more detail about the technology following its launch yesterday.

The new MacBook Pros from Apple are the first devices to come to market sporting a Thunderbolt port. Intel has not been quick to come to market with this technology (it was first announced in mid-2009) and even now the technology which was to be based on fibre optics, comes to market with a copper cord and a maximum speed of 10Gbps. Don’t get us wrong, 10Gbps transfer rates are pretty spectacular but why espouse the benefits of light and then keep us in the dark.

Following Apple’s launch yesterday, Intel has been giving us some more details about the technology and how it works. The Thunderbolt port looks for all the world like the Mini DisplayPort it replaced on the Macs. As you can see from the image above, Thunderbolt consists of a brand-new chip, and a cord, which allows devices to pipe two data streams simultaneously – in both directions – over a single cable primarily using PCI Express x4 for data and DisplayPort for video. Intel has said that while the Thunderbolt chip will be required to use the technology, it can be used with non-Intel chipsets. The technology allows for up to three metre long cables (limited due to the copper), but since the technology also supports daisy-chaining this could be less of an issue.

So what of the new-born USB 3.0? Intel has revealed that it supports this standard: “Intel fully supports USB 3 and plans to integrate it in the future," the company's Jason Ziller told Engadget. Whether this happens because USB 3.0 has taken the market by storm, or vice versa, we will have to wait and see. Intel has also said the Thunderbolt will be backwards and forward compatible depending on the cable used. Intel also promised to roll out the fiibre optic technology soon, with it faster speeds.

For now we will be looking for PC and peripheral manufacturers announcing Thunderbolt-compatible devices in the coming weeks and months with LaCie and Promise the only companies declaring their intentions so far.

Source: Engadget

simonm

February 26, 2011, 12:51 am

Sure Intel 'fully supports USB 3'.





The only reason for releasing this half-baked 'copper compromise' version of their new optical interconnect is because the original design was taking too long to ready for prime time and they needed to get something out before USB 3 achieved critical mass. In order to stave off a nightmarish future where peripheral manufacturers didn't have to pay an Intel tax on every item they shipped.





Presumably waiting for LightPeak (and then getting impatient and falling back to good old copper) is also the reason Intel was dragging their heels on USB 3 motherboards.





And what does "backwards and forwards compatible depending on the cable used" even mean? Does it mean that a future optical version won't be compatible with the copper version, or in fact will it, then all optical Thunderbolt devices will have to implement optical and copper signaling? Messy.

chessnuts

February 26, 2011, 3:05 am

Will Thunderbolt be the one port to rule them all? Will it also replace audio ports such as RCA, 3.5mm jack, and optical?

Gk.pm

February 26, 2011, 3:12 am

@simonm


That's unfounded nonsense. USB 3 implementers also have to pay a licence to Intel..


It actually makes sense to have both a copper (powered) and fibre (unpowered) ports "talking" the same protocol.





But you're right about one thing, Intel did want this out before USB3 achieved critical mass, otherwise they would have an harder time convincing people to use Thunderbolt and then it might not have reached the interest it deserves.





I mean this thing absolutely rocks. PCIe 4x on an external cable, It's mindblowing, possibilities are endeless..

simonm

February 26, 2011, 4:26 am

@Gk.pm





Sorry. If you're right I have misunderstood Intel's USB 3.0 licensing position, expressed by them as "a no-royalty licensing obligation (basically: free, gratis, unpaid, zero dollars, free of charge, at no cost, on the house)" here:





http://blogs.intel.com/technol...





This is further backed-up by the repeated use of the term "royalty-free" in the official USB 3.0 Adopters Agreement:





http://www.usb.org/developers/...





As far as I know they aren't collecting royalties, and they aren't manufacturing the silicon either.





But I'm not a lawyer, heaven forbid... do you actually know they are {fair enough, if so: I unreservedly apologize and please substitute "an Intel tax" with "so much Intel tax" in my earlier post!} or was your cry of "nonsense" a mite random?

Gk.pm

February 26, 2011, 6:14 pm

@simonm





First I must apologise for making such a strong dismissive statement, it's just that LighPeak is such a game changing technology. For the first time mobile devices can have access to true high speed, low latency, and almost unlimited expansion. Dismissing it as just an Intel royalties play is so unfairly demeaning I actually got a bit angry :-)





As for the royalties, it's such incredibly complex area.. From my understanding the royalty-free licensing you referred to is only for xHCI, the interface a host USB controller must present to the operating system. The blog post you mentioned confirms this when it says: "Think of the host controller spec as a &#8216Dummies Guide&#8217 to building a USB 3.0 compatible piece of silicon; it is NOT the USB 3.0 specification itself. "





Intel was almost forced to release this for free due to a range of issues including threats from AMD and NVDIA to fork the platform into two competing interfaces, a bad experience for Intel already back in USB 1.1. I also suspect there was some antitrust issues involved and considerations to the other litigation Intel was experiencing with those companies.





However this royalty-free portion is just a small part of an actual USB implementation. There's a sea of patents covering almost every other aspect: how the bus operates, how chips interface to PCI and PCIe, maybe even connector designs, etc. The complexity of navigating all this shows when only one player, Synopsys, has 85% of the USB 'cores' (industry term for implementatable design) intellectual property market. Even big players like Samsung choose to licence Synopsys USB cores rather than go at it themselves.





Of course Synopsys are good buddies with Intel, with a huge range of undisclosed agreements covering diverse areas (they do compete in some others). They are featured in almost all Intel's USB industry events, an exposure that surely does not come free...





So, apart from the xHCI nothing else, that I'm aware of, in USB's world is free, and portions of those payments certainly drip back to Intel.





Yes, it's possible that Thunderbolt may attract higher royalties for them, although I haven't heard anything about licensing terms yet. We do know that the connector was already royalty-free from Apple and the software interface layer (same level as xHCI) is based on PCI so again royalty-free.





But hey if it lets me plug external graphics cards or computing GPUs, run two monitors, have external truly high-speed USB 2.0 hubs (not 8 ports sharing the bandwidth of 1), high speed external RAIDs, interconnect machines with so much bandwidth and low latency it's like they were a single unit, high speed video capture, and support all this on devices the size of macbook airs or even tablets I think they deserve their 10 or 20p :-)

simonm

February 26, 2011, 11:45 pm

@Gk.pm





I did see that the blog went to some lengths to distinguish what Intel was doing from the specification as such.





But doesn't the USB 3.0 Adopters Agreement on usb.org imply that the whole deal is royalty free? To me it seems to, but I do realize these licensing questions are full of mantraps for the unwary.





I *nearly* think it's very exciting to have PCIe 4x on a standard external port. But then I realize I haven't added a PCI card to my desktop for many years - USB 2 delivers almost every reasonable peripheral except graphics, and with a speedbump to USB 3 this would have continued into the foreseeable future.





And I don't think most users care about external graphics on laptops. Obviously there are exceptions among the kind of people who buy high-end gaming laptops, etc. who care a great deal, but this is a tiny fraction of the PC market as a whole (this segment may be overrepresented on TR!).





So, yes, I do see this as a most unhelpful move by Intel, who were part of the USB 3 committee for goodness sake, but clearly rather grudgingly as they had something better in the pipeline that they weren't going to share (although in the event the pipeline didn't run smoothly - I hear the bill of materials for the planned optical version was going to be $40... I dread to think what this would have translated to in terms of retail price; and some sources claim it never worked reliably).





We now have the kind of fragmentation that is almost certain to slow the adoption of either standard by peripheral manufacturers. Brilliant. Also it would not be surprising if technically literate users were at least a little inclined to put off high-end laptop purchases until this resolved... unless they would be content to find their expansion options on the fast track to obsolescence.





Fundamentally I do think this is borne of Intel's understandable desire to enjoy a monopoly on the interconnect silicon, and I don't think the advantages to the average consumer {who doesn't routinely create RAID arrays of 8 monster-speed drives for the purposes of a press demonstration} of Thunderbolt over USB 3 are enough that we should celebrate the fragmentation.

Gk.pm

February 27, 2011, 6:18 pm

@simonm





As you yourself said USB2 delivers almost everything a "normal" user needs, so there's little point for USB3.If we're going to have a bus for power users then we might as well do it right and throw away the pile of rubbish that USB has been based on from the start.





I don't see any big deal with fragmentation, for sure USB2 ports will continue to be around for many many years and that's what matters for the great majority of users. If Intel is really not supporting USB3, and not integrating it into its chipsets, then I'd say there's really no question about which higher speed bus will be around in the future.





Good riddance too!

simonm

February 28, 2011, 1:20 pm

@Gk.pm





"... no question about which higher speed bus will be around in the future". So that's the forthcoming optical version of Thunderbolt then? ;)





Although there will be forwards and backwards compatibility with the copper version, apparently. Somehow.





I don't think Intel has given us enough at this stage to kill USB 3 (and I suspect that the pricing on their interconnect is going to be unashamedly premium... much more than the 10 or 20p you suggest, enough to add noticeably to retail prices of peripherals). A negative spin on this announcement would be that they've helped to create question marks around both standards that could stall adoption in the manner of HD-DVD and Blu-ray.





But I read somewhere that Apple is likely to have an exclusive on Thunderbolt until 2012. So probably the industry will press ahead with USB 3. Thunderbolt could perhaps displace it later as it migrates down from the high end. In time it could be the single interconnect necessary if it were possible (with the flexibility of PCI and with the extra bandwidth I don't see why not) to create an adapter to plug USB 3 devices into Thunderbolt.

comments powered by Disqus