Intel Demonstrates Light Peak Connectivity Standard Running on Laptop

[nuckinfutz]

Hmm, Apple has been pushing its Display Port technology pretty hard, and ignoring other things like HDMI and eSata. How does this fit in with Display Port or FireWire? Does it compete with it? Replace it? I see yet more more dongles and converters in our future.

DisplayPort is a VESA standard not Apple standard. VESA adopted the mini DisplayPort adapter from Apple and made it a standard of DP 1.2.

DisplayPort is compatible with HDMI through adapters and Apple's never going to support eSATA.

Light Peak will likely replace Firewire and potentially other ports but that's years off.

Light Peak rules.

 

[Reed Rothchild]

Light Peak and SSD would be a big draw initially I suspect. The problem with the video interconnects is that you'd need some sort of conversion device to change the optical signal to electrical. It obviously already exists based on the demos, but seems a bit kludgey until monitors can take a direct light peak connection. In fact pretty much all external devices currently would need that conversion.

Hopefully there won't be an onerous royalty charge to implement this on devices, not that such charges prevented HDMI from becoming dominant. With the reports that Apple actually came up with this idea and brought it to Intel it seems promising that it'll appear on their kit sooner than others. They'll run with anything if it saves a few cents per device .

 

[dba7dba]

Hmm, Apple has been pushing its Display Port technology pretty hard, and ignoring other things like HDMI and eSata. How does this fit in with Display Port or FireWire? Does it compete with it? Replace it? I see yet more more dongles and converters in our future.

It's said that Apple was the one that really pushed/asked Intel to push out Light Peak. Light Peak really goes well with what Apple's about, simple UI.

 

[Eric S.]

This would be great. I'd buy a new Mac for sure, when it's built in to them.

And what devices will you connect to it?

 

[bug67]

"...rumors that Apple played a role in the development of the technology, many observers expect Light Peak to quickly make its way into Macs..."

Kinda like Blu-Ray?

Actually, I really hope to see this technology soon. It is sorely needed.

 

[VenusianSky]

I'm in the same boat. I'll be very mad if this isn't the case. My mac pro is practically brandnew.


On the other hand, with the speed increasing drammatically, how will the hard-drive handle it? Pardon my ignorance, but as far as I know, the common speed for a hard drive is 3gb/s while Light Peak begins at 10gb/s. Will the lightbeam stop and wait for the hard drive?

3Gb/s is the bandwidth of SATA (well there is also 6 Gb/s SATA). There are no single SATA HDD or SSD drives, that I am aware of, that can even transfer at 3Gb/s. 3Gb/s does allow for several drives to transfer with little bottleneck however. In other words, the disk speed will always be the bottleneck to performance, rather the interface.

 

[nuckinfutz]

Kinda like Blu-Ray?

Actually, I really hope to see this technology soon. It is sorely needed.

Blu-ray's a bag of hurt.

 

[satcomer]

I can't wait! I sure hope this will make it's way as a PCI board that can plug-in my Mac Pro. Just imagine having a Mac connected to a cable box via Light Peak.

 

[nuckinfutz]

3Gb/s is the bandwidth of SATA (well there is also 6 Gb/s SATA). There are no single SATA HDD or SSD drives, that I am aware of, that can even transfer at 3Gb/s. 3Gb/s does allow for several drives to transfer with little bottleneck however. In other words, the disk speed will always be the bottleneck to performance, rather the interface.

Wrong

Clearly shows that 3Gbps SATA bus limits the fastest SSD drive by almost 30%.

 

[99John]

What's the Killer App?

So help me understand - suppose lightpeak starts being built into Macs (either this year or, probably more likely, next year). You now have an iMac, say, that can accept a lightpeak signal. If I have an SD card or hard drive full of HD video that I want to copy to my iMac, how will lightpeak help me? What else would you need to really take advantage of lightpeak (since SD cards and camera hard drives don't have lightpeak yet)? I'm just not seeing where this really helps me with the one task that seems like the next big thing for computers - moving tons of HD video around. Thoughts?

 

[ValSalva]

This is why there's no USB 3.0 on the new MBPs. Of course there are few chipsets that support USB 3.0 and there aren't many devices that are USB 3.0 but Apple could have taken the lead much like they did with FireWire and USB 1.0. Much like they probably will with Light Peak.

Mac Pros will probably be phased out by then though as Apple is actively killing that market. Yes that's frustration speaking.

 

[SeanMcg]

So help me understand - suppose lightpeak starts being built into Macs (either this year or, probably more likely, next year). You now have an iMac, say, that can accept a lightpeak signal. If I have an SD card or hard drive full of HD video that I want to copy to my iMac, how will lightpeak help me? What else would you need to really take advantage of lightpeak (since SD cards and camera hard drives don't have lightpeak yet)? I'm just not seeing where this really helps me with the one task that seems like the next big thing for computers - moving tons of HD video around. Thoughts?

Or "how's this USB thingy going to help me when I have all these serial mice with the round plugs?" You have to start somewhere, and I don't doubt that this will gain traction.

From MR's original article on LightPeak

The fact that various types of data including High Definition displays can be driven through the connector has led to some speculation that Apple could incorporate such a connector into their rumored tablet.

iPad 2.0 anyone Let's keep the tablet rumors alive !!!

 

[iketeru]

it remains to be seen whether 10gb/s is with glass fibers, or with plastic. glass fibers would give more bandwidth, but good luck marketing it to people who don't want to have to worry about bending them past the point of breakage.

 

[nuckinfutz]

So help me understand - suppose lightpeak starts being built into Macs (either this year or, probably more likely, next year). You now have an iMac, say, that can accept a lightpeak signal. If I have an SD card or hard drive full of HD video that I want to copy to my iMac, how will lightpeak help me? What else would you need to really take advantage of lightpeak (since SD cards and camera hard drives don't have lightpeak yet)? I'm just not seeing where this really helps me with the one task that seems like the next big thing for computers - moving tons of HD video around. Thoughts?

Here are some scenarios.

You have large storage device say a Drobo 8 bay array with 6 TB of data. This array may be in your basement and you hook up to it over Light Peak at speeds that are faster than if the drive was sitting your in Mac.

another:

Light Peak carries video signals. Imagine having just a monitor and speakers on your desk. The actual computer is in another room connected to Light Peak. You're accessing your storage, ethernet devices and video display data all over Light Peak.

Since Light Peak is multi protocol and will scale to 100GBps bidirectional it's only a matter of time before the thought of a computer making ANY noise is unacceptable. Your computer simply will not be in the same room with you.

With Light Peak you need more bandwidth ....you can route multiple ports. The advantage over copper cables is this. We're talking about light...it moves faster than anything and this means a Light Peak cable should offer latency that no copper cable can match.

Light Peak or rather optical technology is certainly a game changer.

 

[jayducharme]

This is really exciting news. Intel had previously stated that they were planning to implement Light Peak sometime this year. So it looks like they'll make their target date. This is going to be an amazing development. The irony is that moving vast amounts of data off external devices will decrease with time as more people move to that ethereal cloud. But it'll be nice to have just one port and cable that can work for all peripherals.

 

[bug67]

Blu-ray's a bag of hurt.

How so? It works awesomly on every NON-Mac computer I have.

 

[deconstruct60]

The only way that can happen is if Intel license light peak away cheaply or open it up to a forum, so everyone can implement it.

They already tried to previously and it failed. At one point there was a push that USB 3.0 leverage light fiber. There was no buy-in for that with the peripheral folks since this would drive up unit costs. Even if the licensing is low lasers don't come for free. Really good, bendable, "user usage proof", fiber costs more than than copper. Since Lightpeak doesn't multiple protocols the encoder/decoder is going to be more expensive than one that only works for a single protocol (so the electronics costs more).
It is nice for Intel and Apple , who have relatively large profit margins, to buy into increased costs. The key question is why folks without large margins would buy into this. They didn't for fiber-USB.30 . I am not sure what is different now for those folks.


It makes zero sense to use several different lightpeak connections to several different devices each of which are talking a single protocol back to the central computer in the system. The hub-and-spoke USB model is goofy for a multiprotocol conduit. Unless aggregating multiple connections/protocols onto the fiber that is just a more expensive way of establishing the connection. In contrast, a docking port connector or the old Apple Display Connector, ADC, (http://en.wikipedia.org/wiki/Apple_Display_Connector) is makes sense. One fiber carries traffic of several different protocols to/from devices. That isn't a USB "replacement" though. It is more like a ADC replacement, only this time put forth as an industry standard (e.g., let Intel shop it around. ) or a industry standard docking connector (again shopped around by Intel to get larger buy-in.)


So for more expensive devices which are aggregators there is a possibility of high price because get higher value (one connector docking port or one connector RAID box ). However, for keyboards , mouse, camera, cellphone, etc. what's the point? USB 2.0 is less expensive and USB 3.0 has high enough bandwidth for the vast majority of those purposes. (getting HD video off a HD camera is a file transfer. Doesn't need multiple protocols.)



Daisy chaining Lightpeak would also run into issues. Double the laser cost in the peripheral and a highbandwith repeater/switching circuits needed to pass through traffic while skimming off traffic for that specific box.

 

[Pachang]

Situation 1) Apple doesn't care about mac pros anymore and won't update them even though all the parts are available until they feel like it

Situation 2) Apple is getting early access to light peak and the reason they are delaying new mac pros is that they are working with intel trying to build light peak into them.

 

[deconstruct60]

Light Peak carries video signals. Imagine having just a monitor and speakers on your desk. The actual computer is in another room connected to Light Peak.

....
The advantage over copper cables is this.

The other advantage if they dump the copper cable is that it will go to the other room. As long as they try to making it a power transmission as well as data transmission connection the length will be limited. This USB wire wrapped with fiber is likely a derivative of the older USB 3.0 fiber work they had lying around. They will kill off the length property of the data fiber gives then if lapse into wanting to support legacy power transmission features.

 

[99John]

Here are some scenarios.

You have large storage device say a Drobo 8 bay array with 6 TB of data. This array may be in your basement and you hook up to it over Light Peak at speeds that are faster than if the drive was sitting your in Mac.
.

Right, but even if the data on the Drobo can theoretically move at 10 gbits per second across the lightpeak cable, it's not actually capable of being read and written at that speed yet, right? At the moment, lightpeak is a really huge hose for piping data, but it seems to me we don't have enough water pressure to make this thing work like it's supposed to, and so transferring 10s of gigabytes is still going to take forever even with lightpeak. When/how does that change? Is it SSD drives that are needed to make lightpeak work like it's supposed to?

 

[deconstruct60]

Right, but even if the data on the Drobo can theoretically move at 10 gbits per second across the lightpeak cable, it's not actually capable of being read and written at that speed yet, right?

High end storage boxes can move at approximately that speed (similar order of magnitude). They cost about as much as a house though so not sure would be keeping it in the basement. ;-)

Current drobo won't because is using SATA network that will tap out long before 10 gbits per second. However, bulk up the CPU , parallelize the device paths more , and go PCI-e flash ram stick and could go faster than current SAS/SATA links. The cost will be high, but can saturate. It is more of a matter can you afford something that can saturate a SAS/SATA link; not so much can it be done.

 

[foobarbaz]

Hmm, Apple has been pushing its Display Port technology pretty hard, and ignoring other things like HDMI and eSata. How does this fit in with Display Port or FireWire? Does it compete with it? Replace it? I see yet more more dongles and converters in our future.

Firewire is nearly dead, and completely buried. eSATA is a transitional technology and will likely be pushed out by USB 3.

That leaves Display Port, which Apple has been pushing indeed:

First of all, displays with optical ports won't be widespread till 2015 at the earliest. DisplayPort will get us there. And simple adapters can make it compatible with VGA, DVI and HDMI. Optical-electrical adapters aren't quite as trivial.

Second of all: Light Peak is just cabling, it's not a single protocol. That means DisplayPort will just be routed through optical cables instead of metallic ones. Ethernet will work the same way. Maybe the Firewire protocol might even make a comeback!

The beauty is that all this is done in software. Each of those Light Peak ports can serve for display, network or peripheral access. A new protocol can even be added with new OS releases. No more being stuck with the wrong ports for your needs!

 

[deconstruct60]

Situation 2) Apple is getting early access to light peak and the reason they are delaying new mac pros is that they are working with intel trying to build light peak into them.

this is a rather dubious move. Intel claimed several years ago that USB 3.0 would be ready in the 2009 timeframe. It is only really rolling out now. Not sure why Lightpeak would roll out any faster. These are demo units on display. Anyone seen any demos of peripherals that natively leverage lightpeak??? Any other silicon design shops announce they will have working lightpeak implementations ( not much of an open standard if Intel is the sole source of the implementation). It is one thing to slap it on a motherboard for a demo. However, if there is nothing (or less than a dozen things) to hook too that is a major blocker.

It is attractive interface that some vendors will buy into eventually. TV and monitor folks for instance. TVs are getting Ethernet, but it is taking time.



Much more likely that this is case 1 where Apple is dragging their feet because it is a non priority to them. It will show up in month or two and have everything could have had a month ago.

 

[nuckinfutz]

How so? It works awesomly on every NON-Mac computer I have.

Steve told me so. I do have Blu-ray hooked to my HDTV and I like it.

Right, but even if the data on the Drobo can theoretically move at 10 gbits per second across the lightpeak cable, it's not actually capable of being read and written at that speed yet, right? At the moment, lightpeak is a really huge hose for piping data, but it seems to me we don't have enough water pressure to make this thing work like it's supposed to, and so transferring 10s of gigabytes is still going to take forever even with lightpeak. When/how does that change? Is it SSD drives that are needed to make lightpeak work like it's supposed to?

No but in this case we already see that the bus (3Gbps SATA) is already limiting the sequential read throughput of SSD. I think you're going to see the same Enterprise theories for utilizing SSD with standard spinning HDD in the next few years. What'll happen is that storage vendors will sell you a 4,5 or 8 bay array and that array will have an additional slot for a fast SSD that will cache some data. Not only will this improve speed for often used files but it will also allow the array to consume very little power and the HDD drives can be spun down.

 

[bug67]

Steve told me so.

Ahh, gotcha!