Doesn't Displayport already have more bandwidth? Could 10Gb/s even run Apple's 2560x1440 cinema display? With Apple as a driving force behind increasing pixel densities (iPhone 4, new MacBook Airs, new iMacs) and increasing pixel counts... does LightPeak make sense if it can't replace Displayport?
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.
Light Peak Coming in Early 2011 With Apple at Forefront?
- Thread starter MacRumors
- Start date
- Sort by reaction score
Wirelessly posted (Mozilla/5.0 (iPhone; U; CPU iPhone OS 4_1 like Mac OS X; en-us) AppleWebKit/532.9 (KHTML, like Gecko) Version/4.0.5 Mobile/8B117 Safari/6531.22.7)
This is great news! People can stop their hand wringing now as I see quick adoption and USB staying around for a long while. USB is a big advantage for low speed and low cost devices.
The quick adoption will come simply due to the huge leap in speed. Some talk indicates 10Gbps each way. That is pretty darn fast, however I'm not to sure this will be replacing video ports either. Bit rates could change dramatically for video if the resolution of monitors doubles soon. So we might see video ports such as Display port sticking around for a while. Either that or Apple implements multiple Lightpeak ports per machine.
Unfortunately there is much speculation about Lightpeak mostly due to Apples silence on the issue. So it is difficult to imagine exactly what their view on the port is. In fact I'm not sure where the "one port to rule them all idea" came from! It doesn't make sense to use such a high speed and likely expensive port as a replacement for USB for example. A USB mouse or track pad simple is to slow to see any advantage with Lightpeak and that can be said for many other devices.
Lightpeak would be excellent for connecting a laptop to a desktop base station. I imagine this is an idea that is at the top of Apples to do list. It would work great but the laptop would still need to have USB and Display port, otherwise you would be traveling with a bunch of dongles and have connectivity problems every where you go.
On the flip side the port would be excellent for desktop and server products connecting those machines to SAN/NAS devices and networking infrastructure. Even better this would make low cost high performance clusters a reality as all you would need is a suitably fast hub/switch. This port could make Apple hardware very interesting to the enterprise. The underdog here would be a Mac Mini like computer with Lightpeak which could lead to very low cost high performance servers.
In between these two extremes I don't see strong initial adoption. Displacing USB for many devices will be very slow or next to impossible due to the economics.
This is great news! People can stop their hand wringing now as I see quick adoption and USB staying around for a long while. USB is a big advantage for low speed and low cost devices.
The quick adoption will come simply due to the huge leap in speed. Some talk indicates 10Gbps each way. That is pretty darn fast, however I'm not to sure this will be replacing video ports either. Bit rates could change dramatically for video if the resolution of monitors doubles soon. So we might see video ports such as Display port sticking around for a while. Either that or Apple implements multiple Lightpeak ports per machine.
Unfortunately there is much speculation about Lightpeak mostly due to Apples silence on the issue. So it is difficult to imagine exactly what their view on the port is. In fact I'm not sure where the "one port to rule them all idea" came from! It doesn't make sense to use such a high speed and likely expensive port as a replacement for USB for example. A USB mouse or track pad simple is to slow to see any advantage with Lightpeak and that can be said for many other devices.
Lightpeak would be excellent for connecting a laptop to a desktop base station. I imagine this is an idea that is at the top of Apples to do list. It would work great but the laptop would still need to have USB and Display port, otherwise you would be traveling with a bunch of dongles and have connectivity problems every where you go.
On the flip side the port would be excellent for desktop and server products connecting those machines to SAN/NAS devices and networking infrastructure. Even better this would make low cost high performance clusters a reality as all you would need is a suitably fast hub/switch. This port could make Apple hardware very interesting to the enterprise. The underdog here would be a Mac Mini like computer with Lightpeak which could lead to very low cost high performance servers.
In between these two extremes I don't see strong initial adoption. Displacing USB for many devices will be very slow or next to impossible due to the economics.
Where have you heard this? In 1999 I had a Dell Inspiron with a 1600x1200 15 inch screen. Now I have a 15 inch MBP with a 1440x900 screen. If anything, pixel density has gone down...
Wrong. The core chipsets and the CPU(s) set the bandwidth boundaries.
You are confusing the external connection media with internal bandwidth. Internally to the computer it is all electronic and no optical. So adding fiber on the outside of the box does not increase the internal throughput. Adding just about anything to the outside of the box does little to increase the internal throughput.
You're joking. There are tons of high end enterprise shops that are running on fiber now for their core backbone. FibreChannel, Infiniband, and Ethernet among others.
Exactly same motivation that LP goes out and gathers and then aggregates lots of lower bandwidth traffic to make it a problem worthwhile of > 5Gbps fiber speeds.
There is no doubt that a PCI-e v2 x4 (or x8) LP card would be useful. The notion though that could plug in 3-4 of cards of that bandwidth and not choke off the
Yes LP has the potential to be cheaper than doing 10-100GB Ethernet (or Inifinband ) using optical connections. The problem has been volume. Those are so expensive, few buy so the price stays high. LP is cheaper because it cuts a few corners that don't matter over the shorter distances.
NAS's, Video devices, etc. Go back into your parents' basement again and read a manual or two.
Nothing prevents you from plugging in a hub to one of four ports. On the other hand, if I'm traveling with my laptop and want to connect to two or three devices (some of which I may just be borrowing, and some of which might be things like an SD card reader that block access to adjacent ports), I don't want to have to carry around a hub as well. People who use a laptop as a desktop replacement and rarely move it may also not want the additional clutter of a hub.
Wirelessly posted (Mozilla/5.0 (iPhone; U; CPU iPhone OS 4_1 like Mac OS X; en-us) AppleWebKit/532.9 (KHTML, like Gecko) Version/4.0.5 Mobile/8B117 Safari/6531.22.7)
Lightpeak makes sense for a number of individual cases. In fact I see it solving many problems.
As to video you are right about the issue with display densities going up. I'm not sure what the upper limit on Display port is at the moment, it doesn't matter though as you note it demands a pretty stiff and continous transfer rate. On top of that you would need to get GPU manufactures on board. Several things combine to cause me to question Lightpeaks ability to initially displace Display port and other tech. Even if there is a video bandwidth issue Apple could simply add another Lightpeak port, it might not be one port to replace them all but rather a couple of ports to replace many.
It is for other reasons that I see success for Lightpeak. This port should usher in very high performance and low cost for neworking, storage attachment, clustering and new or uncommon uses. Consider this, if each new Mac came with two Lightpeak ports you could have one port dedicated as a clustering port and another for storage or external networking. BOOM, Apple instantly becomes the leading hardware manufacture for scientific computing, database servers and other uses where tightly coupled and quickly communicating hardware is required.
When it comes right down to it I don't think anybody at Apple or Intel see Lightpeaks strong suit is the replacement of USB. Honestly it would be a hardsell to put a Gbps port into devices that barely reach kilobytes per second or operate in the low megabytes per second range.
In any event it looks like we will all know soon enough.
tuna said:
Lightpeak makes sense for a number of individual cases. In fact I see it solving many problems.
As to video you are right about the issue with display densities going up. I'm not sure what the upper limit on Display port is at the moment, it doesn't matter though as you note it demands a pretty stiff and continous transfer rate. On top of that you would need to get GPU manufactures on board. Several things combine to cause me to question Lightpeaks ability to initially displace Display port and other tech. Even if there is a video bandwidth issue Apple could simply add another Lightpeak port, it might not be one port to replace them all but rather a couple of ports to replace many.
It is for other reasons that I see success for Lightpeak. This port should usher in very high performance and low cost for neworking, storage attachment, clustering and new or uncommon uses. Consider this, if each new Mac came with two Lightpeak ports you could have one port dedicated as a clustering port and another for storage or external networking. BOOM, Apple instantly becomes the leading hardware manufacture for scientific computing, database servers and other uses where tightly coupled and quickly communicating hardware is required.
When it comes right down to it I don't think anybody at Apple or Intel see Lightpeaks strong suit is the replacement of USB. Honestly it would be a hardsell to put a Gbps port into devices that barely reach kilobytes per second or operate in the low megabytes per second range.
In any event it looks like we will all know soon enough.
If LightPeak will be released as soon as next year, our IT departement will be furious because I will surely mess around with new adaptors to plug into the network as soon as new devices with this interface are around.
I don't mind having new cables and stuff. Hardware is constantly changing. And there is almost always a low-cost solution to make old hardware compatible to new one. The idea of having just one type of cable is catching, but there will be a transition period of some years, so no haste, no hype, buy some adaptors and go with the time.
I remember when USB 1 came out. People were buying hubs desperately. We'll see how this evolves.
I don't mind having new cables and stuff. Hardware is constantly changing. And there is almost always a low-cost solution to make old hardware compatible to new one. The idea of having just one type of cable is catching, but there will be a transition period of some years, so no haste, no hype, buy some adaptors and go with the time.
I remember when USB 1 came out. People were buying hubs desperately. We'll see how this evolves.
Depends on the type of RAID you're using, but yes, if you used two hard drives that can't saturate a 10Gb connection by themselves, in a RAID 1 i think it is (not sure which is the performance increasing RAID) then you probably would be able to see huge performance increases. Hardware manufacturers could include a dual-drive product in a single enclosure for this exact purpose.
how come dual drives, and say tri drives, aren't more common? Hard Drives are huge bottle necks, in performance, and SSDs are super expensive
Well, they take up room, and the average user doesn't need the reliability or performance of a RAID, or the extra capacity of multiple hard drives. If hardware manufacturers made this, I'd buy it. Imagine a dual-SSD drive running over a single LP cable. The read/write speed would be pretty insane!
This is why Jobs and company are no too keen on USB 3.0. Let the games being.
BTW, niche market. USB 3.0 to LightPeak converter module. Get coding now.
But Apple loves connectors that elimate wires.
http://en.wikipedia.org/wiki/Apple_Display_Connector
LP is a huge excuse to bring the ADC (or something very close to it ) back. Apple already using mini-display port ( as opposed to the regular one).
Do you really think Apple (and Jobs ) really likes :
http://www.apple.com/displays/
When they could cut that down to a two headed hydra instead of three. One just for MagSafe and LP for the display and USB. If they can find some excuse to make it just one connection they will. Actually can if just let folks use the power brick they already own instead of selling another one inside the display.
As far as being "future proof". Light Peak isn't buying that. There is very little that says the 1st generation LP controllers are going to be able to handle future generation traffic.
The display port will be non standard. Since when didn't Apple make you buy a dongle to connect your standard display to their machines ? It has been a couple of years. What makes you think they are going to change now?
there is a limit to how many LP ports you can add without trottling the bandwidth. They are going to be expensive also ( in terms of both relative cost and in internal space consumed. )
It would be nice if the big PC vendors all standardize on a standard docking port. Besides the power, the rest of the standard connectdions could all be run through this one port. Folks wouldn't have to buy proprietary docking stations for each model.
But that would only make sense for the consumers........
LP on the outside doesn't make the insides faster.
I guess when you put a more high power CPU and discrete GPU (for OpenCL horsepower) in there you could make the disks entirely out of the box. However, it doesn't make lots of sense to put that in a milled aluminum box with one fan.
However, can perhaps see Apple nuking the FW and a few USB ports and giving you just a LP option for fast I/O. That wouldn't make the Mini high performance. The external disk box likely won't be inexpensive either.
LP is going to come to peripherals. Just not inexpensive ones. They are going to be products that tend to cost more than USB (2.0 and 3.0 ) ones. They will look inexpensive next to the high end enterprise stuff though (e.g., $800-1000 SAN box compared to a $2000-10,000 one. )
I know!!
or atleast 2 or 3 affordable sata hard drives at 7200 rpms.
I don't really care about storage but the performance is what I want
LightPeak
I did work since mid-2007 on Light Peak ideas. The project is much older and comes from Intels labs. Intel wanted to replace the internal copper/silver/gold based bus systems/conductors (like you find them on so called motherboards), because the required bandwith in the future could not be achieved with the old electron based technology. Photons are not dependend on the capacitance, inductance or resistance of a metal conductor. This is the idea behind Light Bus, or later the easier and external version called Light Peak. Intel works already on Light Bus under a different code name. What that means is, that you will see motherboards without copper/silver/gold conductors in the next 10-20 years.
I did work since mid-2007 on Light Peak ideas. The project is much older and comes from Intels labs. Intel wanted to replace the internal copper/silver/gold based bus systems/conductors (like you find them on so called motherboards), because the required bandwith in the future could not be achieved with the old electron based technology. Photons are not dependend on the capacitance, inductance or resistance of a metal conductor. This is the idea behind Light Bus, or later the easier and external version called Light Peak. Intel works already on Light Bus under a different code name. What that means is, that you will see motherboards without copper/silver/gold conductors in the next 10-20 years.
The Standards Are The Problem, Not The Chipset/CPU Manufacturers
No, the physical properties like the capacitance, the inductance and the resistance of a conductor limit the bandwith. The silizium implements just some standard protocols (different SATA protocols, for example), which have defined bandwith limits. These limits are part of the standards and are not a problem on the silizium level.
No, the physical properties like the capacitance, the inductance and the resistance of a conductor limit the bandwith. The silizium implements just some standard protocols (different SATA protocols, for example), which have defined bandwith limits. These limits are part of the standards and are not a problem on the silizium level.
You are pretty much ignoring the established 10+Gbps Ethernet and Infiniband solutions that are already out there. Therer are already high speed fiber, low latency interconnects out there for use in scientific computing and DB servers. Those have existed for years now.
LP isn't going to move them out of the way without tons of additional help. Furthermore if sell more optical transceivers you will also lower Ethernet and Infiniband prices too. LP's transceivers are not unique to the protocol. Ethernet and Inifinband could use them too if just issue new standard to the fiber being used.
There are going to be some folks who couldn't afford to buy SAN who will get "good enough" LP ones now. It is going to grow into new submarkets which were locked out on price, but not necessarily take over the markets that already existed.
Additionally it is unclear LP will work for clustering. Most of examples shown so far are hub-spoke topology problems. Hooking display to computer. Hooking USB to a computer. Hooking PCI slot to a computer.
Notice the non clustering elemetns to those problems ?
LP could be a generic physically connection transport oriented solution. It just moves bits from one box to another. That means you'd need a stack of something else to solve the clustering issues. The fact you can't get a Inifinband card to put into a Mac Pro say alot whether this clustering support stack (which also has to be layered and developer) is going to show up.
no. the cpu and the devices to which the bus connects only can handle so much data per second (i.e. bandwidth). Even if data transfer operates at the speed of light, that just eliminates latency. If everyone could transport immediately to the genius bar, we'd get there faster, but once we arrived we'd still have to wait our turn.
That's kind of harsh isn't it? I mean, even if LP comes next year (unlikely), it won't be ubiquitous before at least several years later. When that time comes, you'll be looking at a new Mac anyway.
well if it's really that fast, and there are hard drives and flash drives out that support it, then sure i'm up for it
Chances are than any device fast enough to utilize LightPeak will be too power hungry to be bus powered. No one's going to be making light peak 2.5" drives. We're talking about RAID5 boxes here.