Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple and Intel Collaborate on Next Generation 'Light Peak' Connectors?
- Thread starter MacRumors
- Start date
- Sort by reaction score
Great idea
The demo video said 10gb/s symmetrical (each direction) but that doesn't take into account protocol overhead. For 1000 base-T ethernet (gigabit ethernet) the protocol overhead is nearly half the theoretical maximum (1gb/s) so you only end up transferring files at around 500mb/s ... still not shabby.
My point is this; because this tech would be handling transport of multiple protocols simultaneously I wonder what the actual available throughput would be... It might be significantly less than half and possibly get worse as you add more and more disparate protocols.
They said in the youtube clip stated that the display was using 8gb of the 10gb available. I wonder if that included the display protocol overhead (the actual data throughput) or just the video data available to the display...
I have a bunch of questions but its still really cool stuff.
Heard it transfers at 10GB/s. At the moment that's a BluRay movie in 30 Seconds. And plans for it to increase to 100GB/s within the next 10 years.
Also notable is that it can be used in Exchanges to speed up internet connections and phone calls/SMS...
The demo video said 10gb/s symmetrical (each direction) but that doesn't take into account protocol overhead. For 1000 base-T ethernet (gigabit ethernet) the protocol overhead is nearly half the theoretical maximum (1gb/s) so you only end up transferring files at around 500mb/s ... still not shabby.
My point is this; because this tech would be handling transport of multiple protocols simultaneously I wonder what the actual available throughput would be... It might be significantly less than half and possibly get worse as you add more and more disparate protocols.
They said in the youtube clip stated that the display was using 8gb of the 10gb available. I wonder if that included the display protocol overhead (the actual data throughput) or just the video data available to the display...
I have a bunch of questions but its still really cool stuff.
Well, 30fps per eye might be enough for starters, but anyway this shows that 10Gbps isn't very much for very ong time.
You call netbook's storage fast?
I call over sustained 1Gbps fast. To achieve that you used to need 6 number budgets with infiniband or fibre channel. With multilane eSata you can have this with less than tenth of the price.
Also, I can't understand how you can call something that is on the market and widely used and good for what's it for, more "dead" than waporware?
eSata is good for what you originally wanted: local but external storage.
And iScsi works just fine through eSata.
For networked storage 10Gb ethernet will be cheaper than LP and even multilane LP will be oversaturated if there is many displays connected without very expensive LP routers here and there. So for the near future LP will be "one connector for all local" or "one connector not for all networked".
Nope. The device end will still be standards non-compliant like they always are. So you will need a dongle or hub to plug in your USB-1/2/3, FW-400/800/3200, Ethernet-10/100/1000/10000, SCSI, AT, PCI, e-SATA, etc. But that one dongle will have one plug into your computer and your computer will have four such plugs. That's 40 gigabit/sec bidirectional. Your iPhone will only have one of them.
That double HD content they demoed is that Japanese double HD consumer broadcast standard.
http://www.red.com/developer
http://www.red.com/faq/category/epic/
From that link: "EPIC S35 5K uses a 13.8MP sensor, EPIC FF35 6K uses a 24MP sensor, EPIC 645 9K uses a 65MP sensor and EPIC 617 28K uses a 261MP sensor."
Rocketman
I understand the protocol is more like Ethernet than anything else.
Oh - there's huge value in it for Apple stock holders.
Apple will find a way to manipulate this technology to make it more profitable for them and more expensive for us - like using mini display port and it's DL-DVI adaptor that doesn't work.
Sounds cool, would be really helpful to have 1 plug for everything. Only a few concerns:
Power; will it have copper (or whatever material) wires to carry power? I don't want to have to have 2 plugs for a device when all I needed was 1 w/ Firewire/USB.
What if this doesn't catch on? That means 1 extra port on hard drives & whatever. There are some external hard drives out there w/ Firewire, USB & eSATA. We don't need 1 more thing that kinda catches on, but not well enough to get rid of everything else.
How will displays work? Since VGA, DVI, HDMI, go through the video card, would that mean that all connections (hard drive, ethernet, etc.) go through the video card? Or will graphics signals go to the video card and then back out to the Light Peak connectors? I can see a lot of latency in that.
For hard/flash drives, will this work more like firewire or USB? Although USB 2 is theoretically faster than FireWire 400, USB only gets its max speed in bursts while Firewire is more of a steady stream. I've heard that's because Firewire has a chip or something that handles a lot of the I/O stuff (like how some ethernet cards have TCP Offloading engines), but USB doesn't so everything has to go through the CPU. Although cheaper, USB is slower for real use.
Power; will it have copper (or whatever material) wires to carry power? I don't want to have to have 2 plugs for a device when all I needed was 1 w/ Firewire/USB.
What if this doesn't catch on? That means 1 extra port on hard drives & whatever. There are some external hard drives out there w/ Firewire, USB & eSATA. We don't need 1 more thing that kinda catches on, but not well enough to get rid of everything else.
How will displays work? Since VGA, DVI, HDMI, go through the video card, would that mean that all connections (hard drive, ethernet, etc.) go through the video card? Or will graphics signals go to the video card and then back out to the Light Peak connectors? I can see a lot of latency in that.
For hard/flash drives, will this work more like firewire or USB? Although USB 2 is theoretically faster than FireWire 400, USB only gets its max speed in bursts while Firewire is more of a steady stream. I've heard that's because Firewire has a chip or something that handles a lot of the I/O stuff (like how some ethernet cards have TCP Offloading engines), but USB doesn't so everything has to go through the CPU. Although cheaper, USB is slower for real use.
I agree, these protocols have quite a bit of overhead. Too bad it's so difficult to totally overhaul all these protocols. Make there as few protocols as possible that work great. Only real difference would be the physical medium the protocol goes over.
Agreed, eSATA is garbage, the connector is flimsy and I challenge you to run it more than 30 feet. SATA inside the computer is fine... mass storage devices can't touch current SATA speeds anyway.
LP looks to be 4 channel fiber! so you could conceivably run a cable longer than a network line with a HUGE jump in throughput... I'm geek'd over this thaaang! A common cable for EVERYTHING...
Imagine running one fiber line to every room of your house. You could keep your computer in the closet and just have a keyboard, mouse and monitor hooked up where you want to use it!
or
You could have 1 insanely fast computer with tons of storage in the basement and use this to bring your own Video on Demand service to every tv and computer in your house!
I WANT IT NOWWWWWWWWWWW!!!
What's really funny (or sad) is all those pro-Microsoft, anti-Apple people talking about how Apple just wants to change things, how USB 2.0 is good enough, how FireWire "never catched on" (yeah right).
If nobody pushed forward we'd still be using floppy drives, parallel and serial ports, CGA graphics, etc.
USB 3.0 was just a little push forward, relatively speaking. But this LightPeak thing seems like the next big step.
Why push technology one small step at a time when you can jump a few decades at once.
If nobody pushed forward we'd still be using floppy drives, parallel and serial ports, CGA graphics, etc.
USB 3.0 was just a little push forward, relatively speaking. But this LightPeak thing seems like the next big step.
Why push technology one small step at a time when you can jump a few decades at once.
too correct, sir!
If somehow they increased the throughput to 100gb/s the protocol overhead would be much less of an issue.
I wonder what kind of performance you would see if fiber was used as the interconnect for the video card.. or the front side bus.. Imagine a hybrid fiber/copper motherboard.... now that's startrek
Dude I think you better check your tin foil hat for holes, you know for just in case.
What's really funny (or sad) is all those pro-Microsoft, anti-Apple people talking about how Apple just wants to change things, how USB 2.0 is good enough, how FireWire "never catched on" (yeah right).
If nobody pushed forward we'd still be using floppy drives, parallel and serial ports, CGA graphics, etc.
USB 3.0 was just a little push forward, relatively speaking. But this LightPeak thing seems like the next big step.
Why push technology one small step at a time when you can jump a few decades at once.
UMMM..... What?
I've never heard anyone say they're happy about settling with old tech. Who are you quoting exactly?
Everyone wants more speed... It's why computer technology evolves every couple months. Progress happens in fits and starts but I've never heard anyone say I want my CPU slower or I'm happy with 512mb of RAM and a parallel cable.
I'm pro Microsoft and pro Apple... Competition breeds progress and no matter which side of the fence you sit on EVERYONE wants more, better, faster.
Also, Firewire didn't exactly catch on like USB. This was partly due to the licensing fees Apple was charging... Firewire was and still is faster than USB AND has a more rugged connector but the battle between these 2 similar transport protocols was decidedly won by USB....
Cheaper prices... That's what the masses want. Every single computer has a USB port, the same can not be said for FireWire... Heck, apple switched the iPods off FireWire generations ago.
This is years away. While the technology works the human element just isn't there. Getting all the CEO egos to adopt this and then put out the computers and the peripherals at roughly the same time just isn't going to happen. Look h ow long the Blu Ray vs HD took. Look at the wireless spec 801.11n... how many years in draft?
No genius... I wouldn't have been impressed if Ballmer (Incidentally, not an electrical engineer) invented it.
There are compromises here. The video from the Intel lab showed a telecom switch, which if I recall, the 1Gbps switches used to cost around $40,000 ten years ago, back when I was installing fiber.
The problem with being light travel only, means expensive cabling to ensure the light is refracted along the length of the cable properly... watching the connector ends to ensure it is cut correctly, and unlike CAT5, you won't be seeing easy to install connectors with length of cable.
Because light is carried along the path, any accessory on the other end will need to be powered externally (IE, the plethora of wires under your desk will grow, or stay the same) IF copper was added to the wire/connector, expect price of cabling to remain high.
I'd expect this advancement as a port for use to replace FW800, display or network connectivity only. Essentially a highspeed interconnect between powered devices. Sounds eerily like fiberchannel networking, which hasn't exactly taken off on the desktop/portable scale due to cost.
I say that we will only need 2 kinds of cables in the future. One port like the LP that has wicked fast data transfers and can be used with almost anything. Use this for hard drives, display connectors and the like.
We also need something USB like where as the data isn't as fast but it has power going through it. We would use keyboards, mice, and other usb things that need power.
If they could stick the wires for power into the LP without hindering it much then That wold be all that we will ever need (at least for the next few decades)
We also need something USB like where as the data isn't as fast but it has power going through it. We would use keyboards, mice, and other usb things that need power.
If they could stick the wires for power into the LP without hindering it much then That wold be all that we will ever need (at least for the next few decades)
Dude, WHAT! we use AC power because DC loses soooooooo much juice over distance...
I had to put the inverter for my wind generator in the attic because I was losing too much juice with it in the basement.
There aren't gonna be copper leads running with the fiber it's not economical. The cables would be immensely expensive and face it, they're not using this to supplement USB... this is inevitably for use from powered device to powered device (like a monitor or external HDD or a network interface for your computer..) This will most likely connect devices that have their own power supplies...
According to intel it will be on production machines 2010.
It is up to vedors if they want to jump on the bandwagon at that time.
Some will,some wont.
Hopefully apple is with the early adopters.
I don't see any problem with using factory made cables. Does still somebody connect rj45 connectors to their cables by themselves?
It has been said in this thread quite many times that LP will include copper for power. And having pair of copper in the cable won't raise the cost notably.
This might be the first time we get low price optical connecion since toslink.
This time it's only 7000 times faster.
Low price of course only if it gets mass adopted...
This is an Apple effort, implemented by Intel. It should be in the Fall 2010 iteration of MacPro updates. So about a year from now. It should propagate to portable hardware offerings within about a year after that. Presumably handtop devices will also selectively see it, probably as the primary docking option in conjunction with a power channel.
A couple of recent stories that associates with this.
There is a new much larger packet protocol being defined that works on high bandwidth backbones.
100gigabit switching schemes are being deployed now to deal with very high 10gb ethernet services.
The 100m (328 foot) range of this cabling scheme suits itself to cable TV companies using telephone poles and underground utility schemes.
Video protocols:
http://en.wikipedia.org/wiki/High-definition_video
720p 1280x720 158 Mbps (HD)
1080i 1440x1080 266 Mbps (HD)
2160p 3840x2160 1424 Mbps (quad HD)
4K 4096x2160 1424 Mbps (Theatrical)
Rocketman
I'm not sure what protocol overhead you are calculating, as it relates to Ethernet, but typically this refers to the Ethernet plus TCP/IP overhead (and does not include any additional overhead as part of the application layer). If this is the case, your calculation is way off. You can safely assume around 6-8% overhead but it can vary depending on some optional fields in the headers. If we are talking gigabit ethernet and you use jumbo frames then the overhead is about 1%.
That being said, your point that theoretical throughput vs. real-world throughput is valid. To me the big question mark is the switching (they called it a router) technology they use and what if any scheduling they'll do do arbitrate which ports get what percentage of the available bandwidth.
That looks promising. I'm curious as to how well it would hold up to repeated, long-term bending/unbending from portable use.
--
jeeze people complained when they had to give up their IBM Selectric typewriters for a monitor KB and mouse
This is far closer than we think-copper is at its limit and will very soon become VERY expensive; China is hoarding all it can get
This IS the future standard why? FAST.
and like Classic, Mac will eventually drop support for the rest of the mess;
Like anything else this will be ubiquitous and the costs will drop quickly. Like any new product, say, BR players, for instance-now under $100, [although based on VERY poor sales- DVD Q1,2, 2009: $5 billion vs $410 million BR-we may see the end of its use for movies soon-and its likely to get worse in this economy-and with a billion still breathing/eating DVD players out there- hell-even DDL/VOD sales have left BR in the dust: closing in on $1 billion in sales*]
On a pie chart, BR has roughly 4% sales vs 11% DDL/VOD vs 85% DVD; Possible clue: Why still no BR announcement from the coming major iMac refresh, As a storage medium? Again, why has Apple not yet taken the byte? The DDL/VOD numbers hint at CLOUD.
*www.videobusiness.com Marketing
And from an interview with some Intel engineers-this move may make Windows creak and strain to the point of being seen as an 'irrelevant' relic of the 20th century-one reason why Intel became fast friends with APPLE-maybe even on their own went a' courting Steve.
Yes I know im talking old SW vs new HW-which was Intel's whole point
Adaptors will presumably be much more expensive. Also Hubs will be strange and complex because of all the protocols. If an 8 port hub has to also be an 8 port ethernet switch, it'll be a lot more expensive than a USB hub.
If they're aiming to have the attraction of being the ultimate cable over a long period, I think they need to have a 3.5mm connector for mobile devices.
Also, they're going to have trouble with the fact that EU just standardized on Micro USB for charging mobile devices (by law).
If they're aiming to have the attraction of being the ultimate cable over a long period, I think they need to have a 3.5mm connector for mobile devices.
Also, they're going to have trouble with the fact that EU just standardized on Micro USB for charging mobile devices (by law).