Intel bets on Light Peak
Intel's LightPeak technology promises double the throughput of USB 3.0 - so is that why the company hasn't backed the rival standard?
Intel senior fellow Kevin Kahn, speaking at the Intel Developer Forum and quoted by PCWorld, described the company's up-coming optical cable technology as "a logical future successor to USB 3.0," and claimed that Intel was hoping that Light Peak would represent "the last cable you'll ever need."
With Light Peak representing in-house technology, there are certainly a raft of reasons for Intel to be hoping that the standard takes off: not least of which would be that Intel itself wouldn't have to pay royalties on its use, while the company's competitors would be required to send cheques Intel's way if they wanted compatibility with the standard.
Intel claims that it isn't out to steal USB 3.0's lunch, however, but does point out that Light Peak itself supports multiple protocols over a single cable - including the USB standard.
The technology offers consumers advantages over USB, too: due to its optical nature, Light Peak is significantly less fussy about cable lengths than USB; maximum data throughput is significantly increased too, with the technology currently maxing out at 10Gb/s compared to USB 3.0's 4.8Gb/s - and Intel is working on increasing that speed still further.
Kahn claims that Intel expects "both [Light Peak and USB 3.0] to exist together in the market and perhaps on the same platform at the same time," although with plans afoot to launch an industry group to promote the standard as early as next year, we could see Intel chipsets introducing Light Peak support before USB 3.0.
Do you think that optical cabling is the future for high-speed connectivity, or is Intel betting on the wrong horse with Light Peak? Should it stop trying to push its own standard and hurry up with USB 3.0 support in its chipsets? Share your thoughts over in the forums.
Previous Article
61 Comments
Discuss in the forums ReplyOn durability, have you guys played with an optical audio cable? The glass strand in the middle is incredibly flexible because it's so thin, the cladding is usually designed so you can't bend it enough to break the glass inside. If anything it could be more durable than a USB cable.
firewire anyone?
I don't get the negativity. We all understand how fibreoptic broadband is so much better than copper lines, but we don't want the same technology in our peripherals?
Bring it on.
Intel supporting its own tech (not available yet) and puting USB 3.0 in a back seat means USB 2.0 will still be the dominant connection tech in the short term.
On the promise of faster tech (Light Peak) we will be using an old tech (USB 2.0) instead of a "easy to switch to" tech (USB 3.0).
No, but it would limit the cable length.
For "close" devices like portable drives and the likes, it would be fine but probably better (read cheaper) to stick with USB.
At least mobo makers can use 3rd party USB chips, I just wish Intel would offer more PCI-E lanes and generally a bit more bandwidth for them to play with.
Wasnt USB intended for that purpose too?
I smell a big shift to AMD until intel gets around to USB3
lolz
Why would it be any more limiting than USB? For long-length runs you could just disable the charging lines. It wouldn't be difficult to detect how long the cable length is, so you can just disable it over x metres.
Tbh I think the differences are pretty moot - both provide bandwidth well in excess of any practical device that normal folks will use. Once you can run an SSD down it without worry there's no need for extra bandwidth for the foreseeable future.
And I would wholeheartedly agree, if there were things out there that utilise the extra bandwidth. As far as I can see, USB3.0 is fast, but of no benefit over USB2.0 while external devices are limited by other; mechanical; facctors.
However, they could use a non-glass cable that would allow more flexibility than glass.
As to using multiple strands per connector, by enabling more optical sensors of different wavelengths, they could upgrade the bandwidth in the future over the same cable, yay multi-plexing.
Is there really a need for multiple strands, why not just use a couple of different lasers? SS lasers aren't exactly expensive.
One question is, will LightPeak allow us to network stuff together? That'd be its best use imo.
It wouldnt be any more limiting than IN SPEC USB cables, which is 5m for high-speed and 3m for low-speed and guestimated at 3m for USB 3, and even 3m is pretty long and sufficient for most people.
But, optical could allow much longer cables AND also eliminate interference, its why it was included in the USB 3.0 spec as well.
The whole point of frequency multiplexing is that you don't need multiple fibers.
Moriquendi
No, its because the consumer would loose out in the end.
USB 2 will stay the standard, Intel will do everything they can to make sure their technology overtakes USB 3. That way only they stand to profit.
Fiber is expensive, not to make but to buy. Who honestly thinks that these cables will be priced competitively, I expect to see Best Buy HDMI cable prices on these considering who is releasing this.
USB 3 is here, now! What would it take to add it to a chipset, why cant we use the existing technology that works? Why do we have to wait for a closed standard?
They'll have a hard time convincing buisnesses/users to move entirely over to this format as it would obsolete all the usb devices they have up to now. I don't see why they would compete anyhow isn't this aimed at high bandwidth transfers ala eSATA?
I can just see it now: External Hard Drive ships with USB3 as its only form of connectivity since it is widely available and still very functional at 4.8Gbps. Consumer only has LightPeak and USB2 connectivity on PC since Intel is not supporting USB3. Consumer must use USB2 speeds or purchase [likely more expensive due to royalties for Intel] LightPeak-ready enclosure.
The advantages of using a widely accepted (and backwards compatible) standard are great. Sometimes pure speed is not the sole factor in the decision.
oh and showerhead said "obsolete all the usb devices" but in the article it does say that it supports USB without the power...like backward compatible
This is a great move by Intel to push the boundarys beyond the lowest common denominator. I have eSata sockets ont the back of my board but have never used them. Whats one more socket that runs faster than the rest?
Intel are doing nothing wrong here so stop the negativity. This forward thinking should he highly commended.
An add-on chip or board fixes the problem.
There are currently 34 Intel boards listen on Newegg with USB 3.0.
There are currently ZERO Amd motherboards with USB 3.0.
Have fun with your AMD.
Even if that only applies to processors, Intel cannot and will not lock AMD or Nvidia or even add-on board makers out of this as it could create a monopoly situation. Something Intel has been fighting with for many years, governments are just waiting for an miss-step by Intel.
Light Peak with two copper conductors for power should be pretty cool.
If anything, this seems like it's more of an eSata replacement (and networking like Krikket said perhaps). External SSDs are about the only thing that would ever draw over 4.8Gbps in the next few years and you usually only need one so manufacturers can get away with only putting on one then stocking up on USB. Looking at my own computer usage: four powered USB devices, two USB chargers, one USB printer, two front USB ports kept ready for flash drives. None of the four devices would ever need that bandwidth, just mouse and keyboard etc. The two chargers just draw power and use whatever port is most common. Printer, again does not need the bandwidth so it may as well use the very popular USB. Flash drives could possibly use the bandwidth? I doubt it. But then their main use is portable storage that can be used anywhere, and everywhere has USB.
Perhaps others are different, but it seems to me that being common and standardized is often more useful than simply being the fastest. In that regard USB is already here to stay, it's dug itself in so deep with peripherals and chargers for every device imaginable that it won't go away. LightPeak, imo, could at best be an additional support function for a handful of devices.
Also, I'd love to see a standard with round connections so there is none of that "Wrong way up again" thing we all do from time to time. Easily done in a basic optical connection, the 3.5" tosslink is based on the 3.5" plug size.
Or maybe do a deal with apple and use a magsafe connection with a round formfactor. I'm thinking of a laptop with a row of small shallow circular holes along the edge. Each magnestised to take the end of an optical cable that will provide uber data speeds and enough power for the average mobility device. Put any device in any socket.
Alternatively we could just drop the wires all together and use something wireless
Hold on... doesn't Sony want us to do that.
SONY!!!! Agghhhhhh I'd stick with Intel thank you very much
I like the idea personaly as it can only push bandwidth of all the device busses up, and I think that is the only way to improve the GFX to an extent that'll make us do more than raise one eye-brow.
Also, on the power issue, a laser is in itself a form of energy. That is why it's meassured in W/KW. I may be wrong again but surely such a precise power source wouldn't require a load of conditioning circuitry either.
Go on, proove me wrong and I'll just go and sit in the corner and sulk while muttering about how stupid and far fetched I sounded.
You get a Copper Coating on top of the Fibre Cable itself. Which means you are still using 2 Cables inside Lightpeak, but you get Power as well.
Consider this to replace every single connection you have, internally and externally, SATA, USB, Firewire, Etherent... etc....
Not to mention the cable would be even thinner then USB 2.0 ( When you consider USB 3.0 is like Ethernet CAT 6 thick.... that is ridiculous......... )
Optical transmission would not speed up on board buses because they are not limited by long lengths of cable or the need to use a limited number of cables. Optical is overly complex for this purpose hence inefficient examples below:
Digital Signal (send) -> tiny length of copper (with as many channels as necessary for bandwidth) -> Digital Signal (receive)
as opposed to
Digital Signal (send) -> Optical converter chip -> laser source -> Fiber Optic -> photo diode -> Optical converter chip -> Digital Signal (receive)
I know what you're saying but you seem to forget that digital is a sieries of ones and zeros. Why convert at all when light is the same language, one=on, zero=off.
The whole curcuit would only require a very little power as no electricity would need to be pushed(amps). The chips would probably use a solar cell to read the information any way, so it would do the job of recieving the power as well.
If you can power a calculator in a sparcely lit office, then think what you could power with a pure laser(which I believe is in the invisible spectrum range).
It's just up to the peripheral manufacturers to forget about their shares in the power companies and start working together as it requires a total rethink and change of the current standards. That goes for the GFX card munufacturers that believe we need to draw the same power as a hallogen oven(and create the same heat output as well).
Shoot!!
Fail. That'll never hapen will it. :(
Almost all optical communication systems are digital all the way but you still need more components to produce an optical bus than an electrical bus, a laser and most likely a photodiode and a converter chip (to clean up the signal) at the receiving end, this makes it less suitable for consumer level use than an electrical bus. Also, there's no way currently to produce optical fibres in or on a PCB.
Also, the is significantly less power available at the end of an optical fibre than there is to a calculator in an office and calculators are very low power devices, powering peripherals optically simply will not work.
Optical communications don't really have any advantages to a consumer at the moment and they wont until required bandwidths increase beyond the ability of electrical busses. Which may never happen as electrical busses are improving faster thant he devices they're attached to, why have a harddisk capable of ~150-200 MBps connected with a bus capable of 10GBps? especially if you cave to pay more for it.
Optical interconnects do have advantages in other situations where interference is a problem (manufacturing) or there are long distances (telco) to be covered or very high bandwidth is required (network backbones)
Moriquendi
:D
So Light Peak is just the cable, just a pipe to send whatever through, be it USB, SATA, IP etc.
What the hell is wrong with the many garden varieties of fibre that are widely used at the moment? There are plenty of optical technologies at the moment which can pipe 10Gb+ over fibre now. Never mind the impracticalities of fibre over a copper cable. Don't get me wrong, fibre is the best thing since bread, but its advantages over copper are moot at home.
And what's the point of having any more than 2 strands? hmm.
Generally speaking: :?
and the people claiming no need for bandwidth are like the no one ever needs more than 24K of ram type. sure 10GB/s right now might be "too fast" but when sata 1 was launched it was "faster' than any pata drive could push and look how that changed. in 3-5 years a SDD array on a NAS box could push 10GB/S easily.
FIber optic is the future, it's already proven itself for long haul and even last mile to customer houses. Slowly coming into consumer range is next especially due to the advantages in throughput it offers over electrical buses.
You can already replace cat5 with fibre but you'll pay through the teeth for it, I haven't heard yet how Intel intend to keep the price reasonable.
My point was that electronic busses are improving too, in 3-5 years we may very well have usb4 or something similar that can push that much data down a copper wire
Why? Fibres advantages are not well suited to the consumer space, the added cost is not rewarded with added performance. Put simply, fibre in the consumer space is a solution looking for a problem. In the long haul and even in the last mile fibre has advantages, it is currently difficult or impossible to push bandwidth that people will use over those distances over copper but for distances under 100m you can do it with copper and at significantly lower cost.
Maybe electronic busses will hit a bandwidth wall while devices demand more bandwith in which case, if fibre can offer the bandwidth, I'm all for it but that is not the case at the moment.
Moriquendi
I doubt intel is going to have much luck on slowing USB3, it's here and it's going to stay.
I do like the idea of 10Gbs fibre for external storage and networking.
With the amount of data floating around theses days I feel gigabit is now becoming some what of a bottleneck
Most new motherboards come or are about to come out with additional USB3 sockets anyway in addition to the many USB2 sockets and eSata. So why are Intel getting dumped on by the doom sayers here?
This is progress being driven positively forward by Intel so why are you all moaning?
Well Done Intel for taking on another area of the industry and making it better (by a long way!)
Have you handled a USB 3 cable lately? They got 8 conductors in them are have the rigidity of a steel cable. An optical cable is positively flexible in comparison. There are also types of fibers which can take a lot of abuse and very tight corners.
For printers and scanners, it's useless.
For external drives we already have Esata.
For Lan, we already have optical, and 10gigabit is coming soon. Plus, I have to wonder about lag on Lightpeak.
Then as mentioned you have the stiff cable...
It's impractical.
Ask yourself, how badly do most people really need USB3 right now?
How much would it cost Intel to re-engineer another chipset to add USB3?
How much to then do it again to add Lightpeak?
So while you are bitching about Intel (who's already boards have it , making your complaints rather pointless), AMD isn't even on the map. Not that most people have a need for USB3 anyway.
The 5Gbps seems the best choice, even more after factoring that you dont really need 10Gbps for anything.
Your confusing "Intel could" with "Intel is".
That would make sense because Apple is woking on this with Intel. Not sure why they did not mention this joint venture in the article. It is probably powered more by Apple anyway now that I think back on all the useful and useless alternative sockets Apple has had a hand in engineering. They can't seem to help themselves. But hey, all devices including video through one connection could be pretty cool.
The cost of the detector needed is pennies.... just saying.
The plastic should be plenty durable enough.
I don't see power as being a big problem with distance, think power over Ethernet. The problem is that DC can incur pretty hefty losses over longer runs, but that may not be an issue for most practical application.
The real problem is the Intel license on the tech. I think we will see a competing, open optical standard develop and that will gain acceptance because it is royalty free.
LightPeak is still no way to be seen.