Apple pushed Intel to develope 10Gbps "Light Peak"

[faster3200]

In the end, we can't say we know or understand Apple's complete plans for the standard, but what is notable is that the company is pushing for an all-on-one connectivity solution, and pushing hard. That means Cupertino is at least prepared to abandon the standards we now know for a singular solution, and potentially just skip over forthcoming offerings like USB 3.0 (or at least downplay their use). It's not an uncommon move for Apple, but in the past those choices have had major repercussions -- and this pairing is extremely reminiscent of the USB / iMac story (Intel also created that standard, which Apple then popularized). We're going to keep our ears to the ground on this news, but for now, it's an interesting peek behind the curtain, and a possible glimpse into the future of connectivity.

Article: Link!
Video: Link!

[zanzabar]

whats with this guy, apple never popularized USB they hatted it, apple users hatted it, then they implemented late affter it was common every ware else then the sheaple users liked it.

unless its cheap like usb and intel puts it on everything it will go the way of firewire, we could have a standardized 10/100GB eathernet port by now and that could be used for everything but i guess that we cant have that when a new expensive interface could be made for apple (i think that apple is just jumping on the press though)



then optical cables suck for removable device use

[Serra]

I'm not sure I fully "get" Light Peak. I see on Wiki that it says the speeds will start at 10Gb/s - is that actually true? If so, what will the costs be like? Current 10Gb/s infrastructure is hideously expensive... far, far out of what any consumer could possibly afford. If the speeds are lower though, then I guess I'm just not sure why we need yet another standard and another cable... one which quite likely has a lower lifetime (lasers aren't known for longevity, and I don't think LED's could get up to 10Gb/s without using a large field of them, though I could be wrong on that).

Don't get me wrong, if prices are actually low enough for this to be useful and it does start at 10Gb/s then I'm on board.


Edit: Waaaaait a second. I see they're also trying to push for LAN to be done through lightpeak? Okay, now I know they're pushing for too much. Speeds right now aren't held down by the cabling standards, they're held down by processing power at the router/switch node. We have cabling right now that can do 10Gb/s but we're on 1Gb/s because only highly expensive equipment can handle the throughput (let alone the signaling). This is a very serious chink in the "It's a 10Gb/s cable that will do it all!" theory and until I learn more about this implementation I'm a little scared of what will happen if it does get aggressively pushed out.

[lowfat]

whats with this guy, apple never popularized USB they hatted it, apple users hatted it, then they implemented late affter it was common every ware else then the sheaple users liked it.

Not Correct. By 1998 Apple had already released a computer w/ no legacy ports. All it had USB and EE1394.

[Katanai]

I'm not sure I fully "get" Light Peak. I see on Wiki that it says the speeds will start at 10Gb/s - is that actually true? If so, what will the costs be like? Current 10Gb/s infrastructure is hideously expensive... far, far out of what any consumer could possibly afford. If the speeds are lower though, then I guess I'm just not sure why we need yet another standard and another cable... one which quite likely has a lower lifetime (lasers aren't known for longevity, and I don't think LED's could get up to 10Gb/s without using a large field of them, though I could be wrong on that).

Don't get me wrong, if prices are actually low enough for this to be useful and it does start at 10Gb/s then I'm on board.


Edit: Waaaaait a second. I see they're also trying to push for LAN to be done through lightpeak? Okay, now I know they're pushing for too much. Speeds right now aren't held down by the cabling standards, they're held down by processing power at the router/switch node. We have cabling right now that can do 10Gb/s but we're on 1Gb/s because only highly expensive equipment can handle the throughput (let alone the signaling). This is a very serious chink in the "It's a 10Gb/s cable that will do it all!" theory and until I learn more about this implementation I'm a little scared of what will happen if it does get aggressively pushed out.

I'm afraid that you don't really understand what the purpose of this is. It's, put simply, more of a dock port for laptops than anything else. It's meant to connect a laptop to all external devices inclouding routers and internet connections in general. I don't see anything wrong with that...

[stevecs]

@serra: Well, yes and no. Yes, there are asic constraints (memory, et al) that are involved in pushing huge bandwidth, but we do have tbit routers available today if you can afford them which goes to your point of cost. However even in these types of environments we are reaching limits of what can be done with copper. The aggregation points are getting constrained and there is push to convert them to optical switches at chip level themselves. Optics have another very good option which is to increase capacity you can add multiple wavelengths down the same light path (this is how DWDM works).

I actually like this as it opens up commoditisation of this technology (or the potential to) to the end user. It's not needed now for a desktop user, think 10 years from now. This is going to be, unless there's a paradigm shift somewhere, where we're headed. We're already there for enterprise builds. I think Apple is just trying to stake out the territory, nothing wrong with that if it pushes the tech further along.

As for the 'optics suck for removable device use', umm. nope, actually it's very good in that regard. The only real detractors are going to be cost & power transmission to the remote devices.

[Serra]

I'm afraid that you don't really understand the purpose of this is. It's put simply more of a dock port for laptops than anything else. It's meant to connect a laptop to all external devices inclouding routers and internet connections in general. I don't see anything wrong in that...

I understand the purpose, what I'm concerned about is performance and appropriateness.

If it operates under 10Gb/s, then we need to know what speed it actually runs at because there are a lot of purpose-built cables out there that might handle our data faster. If it is hideously expensive, we need to know that too.

As far as what it is "meant to do" - it's meant to be able to connect anything to anything. But you can't just arbitrarily do that. You can't just arbitrarily connect to a switch then to another PC and expect to get 10Gb/s speed - the switch/router has to actually process data at the throughput speed, and I promise you that no consumer technology can do that today. Which brings us back to my question - How Fast Is It?


@steve:I'm not saying that this couldn't be good - I want to see pricing and exact speed specs. As to the copper issue - yes, we're at about the limit of Cat 6. Cat 7 is available though, it's just not popular because there is no reason to use it right now. I'm not against fiber for connecting to things around the home, but I think it's something to approach very, very cautiously. I also know about DWDM, I would be shocked if multiple wavelengths weren't in use to get the 10Gb/s out of a link they plan to sell to customers.

I also have a problem with people pushing optics for home use because:
1. It's not installed in any homes (yet) - maybe get it installed then worry about replacing ports with it
2. If you have a jack issue... man, you have to hire a pro to come fix it, and the fix is not cheap.

Edit: Steve, what are you doing out of Storage? Back to your cave! :P j/k, it's nice to see out out and about.

[Firestrider]

So this is meant to replace FireWire, USB, DisplayPort, VGA, HDMI, DVI, SATA, RCA audio, and CAT5/6 cables?

If so I'm all for it.

But I guess that means I will need a lot of adapters or new peripherals

[Katanai]

I understand the purpose, what I'm concerned about is performance and appropriateness.

If it operates under 10Gb/s, then we need to know what speed it actually runs at because there are a lot of purpose-built cables out there that might handle our data faster. If it is hideously expensive, we need to know that too.

As far as what it is "meant to do" - it's meant to be able to connect anything to anything. But you can't just arbitrarily do that. You can't just arbitrarily connect to a switch then to another PC and expect to get 10Gb/s speed - the switch/router has to actually process data at the throughput speed, and I promise you that no consumer technology can do that today. Which brings us back to my question - How Fast Is It?

So this is meant to replace FireWire, USB, DisplayPort, VGA, HDMI, DVI, SATA, RCA audio, and CAT5/6 cables?

If so I'm all for it.

But I guess that means I will need a lot of adapters or new peripherals

Not really. I'm pretty sure that, at first, the only thing that it will replace is the dock port. Those 10GB are not meant for a single data connection, though they could be used that way with an adapter. That 10GB line will be split to all connections. Let's say 1GB for ethernet, 1GB for HDMI, 1GB for USB ports and so on. If I'm right this could be a good tech for the future and it does indeed fall in line with Apple's minimalistic design philosophy...

[Russian]

I'm not sure I fully "get" Light Peak. I see on Wiki that it says the speeds will start at 10Gb/s - is that actually true? If so, what will the costs be like? Current 10Gb/s infrastructure is hideously expensive... far, far out of what any consumer could possibly afford. If the speeds are lower though, then I guess I'm just not sure why we need yet another standard and another cable... one which quite likely has a lower lifetime (lasers aren't known for longevity, and I don't think LED's could get up to 10Gb/s without using a large field of them, though I could be wrong on that).

Don't get me wrong, if prices are actually low enough for this to be useful and it does start at 10Gb/s then I'm on board.


Edit: Waaaaait a second. I see they're also trying to push for LAN to be done through lightpeak? Okay, now I know they're pushing for too much. Speeds right now aren't held down by the cabling standards, they're held down by processing power at the router/switch node. We have cabling right now that can do 10Gb/s but we're on 1Gb/s because only highly expensive equipment can handle the throughput (let alone the signaling). This is a very serious chink in the "It's a 10Gb/s cable that will do it all!" theory and until I learn more about this implementation I'm a little scared of what will happen if it does get aggressively pushed out.

You mean you actually think apple cares what consumers can afford?

[Serra]

Okay, so I finally just noticed - it does not RUN at 10Gb/s, it has 10Gb/s of bandwidth. Speeds have not been mentioned at all. Thus I would assume for now we'll start low and build up, maybe see some 1Gb/s stuff to replace LAN connectivity eventually etc.

I wonder how they will designate it then. Will we have motherboards with 3x LightPeak-100 cables for 100Mb/s speed, and 2 LightPeak-1000 cables for gigabit speeds (etc)? I can see this being quite a headache for many people as they discover speed mismatches. It will be like USB 1.1 -> 2.0 all over again... repeatedly. Though likely with better speed results.

<snipped out a quote/response that I had misread>

Edit: One last question that I'm wondering if anyone has seen addressed yet - with fiber optics in networking, we tell people to NEVER look at the fiber when a laser is running in it as baaaad things happen. With LED's is not such a big deal. Has anyone heard anything on this one way or another?

Edit II: You can probably tell I'm geeking out on this a bit. My work and favorite hobbies revolve around device interconnects

[zanzabar]

Not Correct. By 1998 Apple had already released a computer w/ no legacy ports. All it had USB and EE1394.

that was my point, apple had hatted usb and had it for comparability (not by default ether) and wanted everything on firewire, then they went for usb and firewire was still better then they killed firewire as the prefered connector when they joined with intel. apple had nothing to do with the adoption or creation of usb


and optical cables for a peripheral is a pain, ive broken so many optical cables for the stereo (or have had them broken, im not sure how many i killed)

[faster3200]

My guess is this is designed to be an all-in-one jack for mobile devices not a USB or Ethernet replacement altogether (for now at least). Apple wants this design because they promote "simplicity" so you can bet they will use it first and probably not just on mobile devices (gotta have the "new" factor).

[Drwho?]

Well, photonic research were started in the beginning 2000's , in the 2nd floor of my building in Santa Clara ... The credits should go to those guys who have the patents about photonic transistors.
Any other claim is bogus, those guys are on it since 2000 ...

Francois

[[XC] Lead Head]

Okay, so I finally just noticed - it does not RUN at 10Gb/s, it has 10Gb/s of bandwidth. Speeds have not been mentioned at all. Thus I would assume for now we'll start low and build up, maybe see some 1Gb/s stuff to replace LAN connectivity eventually etc.

I wonder how they will designate it then. Will we have motherboards with 3x LightPeak-100 cables for 100Mb/s speed, and 2 LightPeak-1000 cables for gigabit speeds (etc)? I can see this being quite a headache for many people as they discover speed mismatches. It will be like USB 1.1 -> 2.0 all over again... repeatedly. Though likely with better speed results.

<snipped out a quote/response that I had misread>

Edit: One last question that I'm wondering if anyone has seen addressed yet - with fiber optics in networking, we tell people to NEVER look at the fiber when a laser is running in it as baaaad things happen. With LED's is not such a big deal. Has anyone heard anything on this one way or another?

Edit II: You can probably tell I'm geeking out on this a bit. My work and favorite hobbies revolve around device interconnects

The lasers used for this type of interconnect would probably be under 5mw, the strength of your average cheap nasty red laser pointer. Their life span is usually spec'd about the same as a good LED, so probably around 40,000+ hours of continuous on. Not pulsing like in a fiber situation. You'd also have to try real hard to get eye damage on low power system like what would be used here.

[saaya]

Well, photonic research were started in the beginning 2000's , in the 2nd floor of my building in Santa Clara ... The credits should go to those guys who have the patents about photonic transistors.
Any other claim is bogus, those guys are on it since 2000 ...

Francois

yes, everything related to photonics has been invented by intel... or stolen or copied from then. hail intel!

please... and you call ME a fanboy...

i dont get whats supposed to be so special about this...
its basically apple doing firewire all over again?
they always wanted a faster standard and didnt mind the increased costs for it... i highly doubt they will skip usb3 though, you always need usb, you cant just stop using it, same for ethernet...

[Darakian]

that was my point, apple had hatted usb and had it for comparability (not by default ether) and wanted everything on firewire, then they went for usb and firewire was still better then they killed firewire as the prefered connector when they joined with intel. apple had nothing to do with the adoption or creation of usb


and optical cables for a peripheral is a pain, ive broken so many optical cables for the stereo (or have had them broken, im not sure how many i killed)

Not true, the first imac was the first computer to standardize usb for mice, keyboard, mass storage, etc.. They may have preferred firewire for the mass storage but they did in large part seed the market with usb ports.

[Serra]

Not true, the first imac was the first computer to standardize usb for mice, keyboard, mass storage, etc.. They may have preferred firewire for the mass storage but they did in large part seed the market with usb ports.

When you say "standardize USB for mice, keyboards, mass storage, etc", do you mean "dropped PS2 support and did not force firewire"? Because I'm pretty sure that USB mice and keyboards existed before the iMac, it was just optional whether you used them or not on PC's (which had USB for a long time).

I would not go so far as to say that a company which caters to ~2% of the market (at the time) was a champion for USB by removing PS2 support long after everyone else had already brought out USB.

[railhan]

It surely won't replace Ethernet. Optical fibers just don't fit in an average Joe's home. For copper cables you need just RJ45 connectors and crimping pliers - no fancy stuff involved. Plus unmanaged Gigabit switches are already cheap enough.

It won't replace SATA as standard for hard drivers.

Maybe .... if its cheap enough, easy to implement, rugged and also if they add power source .... it might replace USB in next 10 years.

[Fuji]

Just throwing this out there. But i think it'd be interesting if it was to be used for inter-cpu or CPU--> other stuff connections.

[Vinas]

Not Correct. By 1998 Apple had already released a computer w/ no legacy ports. All it had USB and EE1394.

Well... Nintendo created the first USB variation for the gameboy. Does that count?

[lutjens]

I personally don't care who initiated it...as long as it works. I think Apple realized that in order to give such an interface a better chance of success, Intel would be the best way to go, with it's dominant market position. And I feel better with Intel doing it...at least it won't end up being some rigidly controlled, proprietary Apple standard. If it were, I can just imagine being told by Apple when you can use it and for what devices (doubtlessly only devices offically approved by Apple would be permitted).

A new standard for everything could be a very good thing, especially considering the very promising ranges with the new interconnect...such a device would be a boon for devices like KVM modules, with everything going over a single connection much like it does now, but providing a great deal more bandwidth than an old 1 Gbps Ethernet cable. But it might end up changing them into nothing more than a fan out device that would split the interface into all the components of a PC. Anyway, it'll be much more simple in any event. Hopefully, one will be able to "team" separate connections together for even faster performance.

I find it somewhat strange that only 10Gbps is being used...I'd argue for an even faster speed as a hedge to ensure that the much faster devices of the future won't be bottlenecked. The maximum bandwidth of the 10 Gbps design would be 1.25 GB/sec, which is very good today, but might be a limiting factor tomorrow. 100 Gbps would be much better...12.5 GB/sec...

Anyway, hopefully this interface arrives sooner, rather than later.

[faster3200]

I find it somewhat strange that only 10Gbps is being used...I'd argue for an even faster speed as a hedge to ensure that the much faster devices of the future won't be bottlenecked. The maximum bandwidth of the 10 Gbps design would be 1.25 GB/sec, which is very good today, but might be a limiting factor tomorrow. 100 Gbps would be much better...12.5 GB/sec...

Did you watch the video? The Intel rep said that it was possible to increase the speed up to 500Gbps in the future. That would certainly be quite amazing.

[m0da]

Here's another, quite informative, article:

http://news.cnet.com/8301-30685_3-10362246-264.html

[saaya]

yeah thats the good thing with optronics, its usually the sender receiver logic that limits, not the cables