Intel Says Fiber Optic Thunderbolt Cables Coming Later This Year

[MattInOz]

It's unfortunate that the Apple T-Bolt Display is so underwhelming. If it had included a couple of 2.5" SSD internal drive bays, USB 3.0, and two or four PM-capable eSATA ports if would be a great docking station. (and the icing on the cake would have been an option for an external graphics card....)

This to me is where some one like nVidia should be coming in to the thunderbolt picture. Get them into making a decent graphics reference design and chipset for 3rd party TB displays and the market starts to change it's colour.

I wouldn't call it a train wreck yet. the Trains not moving fast enough to jump the tracks yet.

 

[hchung]

Show me a 2m copper cable that can handle 2 channels of 10 Gbps full-duplex for much less than $49. There is a lot to the Thunderbolt cable spec, and it's not just the fact that it's an active design. You can read more about it through Apple's patent application: http://www.patentlyapple.com/patent...al-thunderbolt-is-headed-for-ios-devices.html
(note that I totally disagree with Patently Apple's title for this article and their assessment that Thunderbolt would be headed to iOS devices any time soon.)

I think someone beat me to the punch on this one, but price out the QSFP+ connectors for both ends of that $0.80/m cable and you might start to realize how cheap Thunderbolt is.

Quoted for truth. xinu appears to have no idea that pricing out a DSP that works with data at 10 gigabits is way different than at 100 megabits.

Say we decided to look up appropriate DSPs from TI to get a ballpark range.

Texas Instruments C5000 DSP is listed for $2-$10. (your guitar pedal)
Texas Instruments C6000 DSP is listed for $40-$200. (your thunderbolt port)

Throw in a couple of connectors at $17 per end and a pair of C6000s, and you're looking at a whopping $110 for a cable. At that kind of pricing, Apple's paying for your cable when you buy it at $50.

 

[AidenShaw]

You might have read it that way, but I was talking exclusively about TB.

Sorry then, I misread your statement by assuming that you were referring to other Apple interconnects that were deemed failures.

Do you really see the space in the CinemaDisplay for all the ports and bays you describe? There's a point where they're having to significantly alter the product to the benefit of a very few people.

Two rejoinders:

• Making the T-Bolt display a cm thicker doesn't significantly alter the product, especially if it would create an innovative docking station that gives a docked laptop some of the flexibility and power of a mini-tower
• The Imac has lots of stuff inside in a similar form factor

And there's lots of space inside the T-Bolt display:

http://www.ifixit.com/Teardown/Apple-Thunderbolt-Display-Teardown/6525/1

 

[repoman27]

Well mr smart guy tell us? How much those DSP I/O chips costs? A dollar? Two? Ten dollars each?

Do you know how much it costs to manufacture an ARM A5 chip? I do. Do you?
Do you realize, that A5 chip is 1000x more complex than that Intel LightPeak DSP Chip?

I'm going to go ahead and give you a huge benefit of the doubt, and take A5 to mean the Apple A5, an SOC featuring a dual-core ARM Cortex-A9 MPCore CPU and a dual core PowerVR SGX543MP2 GPU, rather than just a plain old ARM Cortex-A5 processor. Do you realize that it is built on the same 45nm process as the Light Ridge Thunderbolt controller, and that the Light Ridge 82524EF/EFL measures 225mm^2 while the A5 SOC is only 122mm^2? That would make the Thunderbolt controller 84% more complex than the A5 and 84% more expensive to manufacture. What's more, a Thunderbolt cable is capable of throughput equivalent to 1/5 of the entire memory bandwidth of the A5 when paired with dual-channel LPDDR2-800 DRAM. Not only can it achieve signaling rates this high, but it can successfully transmit those signals for several meters rather than the few mm that the A5's memory bus is capable of.

Despite your quaint refusal to believe otherwise, designing cables capable of supporting 20Gbps, full-duplex for consumer applications is a non-trivial challenge. In fact, let's look at the short list of what's out there: 4X QDR/FDR InfiniBand, 40GbE, and ePCIe 2.0 x8/x16. That's all I can come up with.

Despite all the hyperbolic statements that seem to crop up on Thunderbolt threads, I rarely see any that actually have a ring of truth to them. So here's one that as far as I can tell really is true: Thunderbolt is hands down the fastest external I/O interface ever to be included by any OEM in a consumer PC.

I have difficulty figuring out why so many people are seemingly opposed to Apple including a radically fast I/O interface in their notebooks and all-in-one PC's. The only thing I can reckon is that some people feel that they would have been better served had Apple devoted their resources to including USB 3.0 in the 2011 Macs instead of waiting a year for Intel to integrate that functionality into their chipsets. I wonder what people will bitch about once all the 2012 Macs arrive and have both Thunderbolt and SuperSpeed USB?

 

[xinu]

I'm going to go ahead and give you a huge benefit of the doubt, and take A5 to mean the Apple A5, an SOC featuring a dual-core ARM Cortex-A9 MPCore CPU and a dual core PowerVR SGX543MP2 GPU, rather than just a plain old ARM Cortex-A5 processor. Do you realize that it is built on the same 45nm process as the Light Ridge Thunderbolt controller, and that the Light Ridge 82524EF/EFL measures 225mm^2 while the A5 SOC is only 122mm^2? That would make the Thunderbolt controller 84% more complex than the A5 and 84% more expensive to manufacture. What's more, a Thunderbolt cable is capable of throughput equivalent to 1/5 of the entire memory bandwidth of the A5 when paired with dual-channel LPDDR2-800 DRAM. Not only can it achieve signaling rates this high, but it can successfully transmit those signals for several meters rather than the few mm that the A5's memory bus is capable of.

Despite your quaint refusal to believe otherwise, designing cables capable of supporting 20Gbps, full-duplex for consumer applications is a non-trivial challenge. In fact, let's look at the short list of what's out there: 4X QDR/FDR InfiniBand, 40GbE, and ePCIe 2.0 x8/x16. That's all I can come up with.

Despite all the hyperbolic statements that seem to crop up on Thunderbolt threads, I rarely see any that actually have a ring of truth to them. So here's one that as far as I can tell really is true: Thunderbolt is hands down the fastest external I/O interface ever to be included by any OEM in a consumer PC.

I have difficulty figuring out why so many people are seemingly opposed to Apple including a radically fast I/O interface in their notebooks and all-in-one PC's. The only thing I can reckon is that some people feel that they would have been better served had Apple devoted their resources to including USB 3.0 in the 2011 Macs instead of waiting a year for Intel to integrate that functionality into their chipsets. I wonder what people will bitch about once all the 2012 Macs arrive and have both Thunderbolt and SuperSpeed USB?

Why they didn't put those chips straight to Mac's and TB-devices? Why so stupid design that the I/O chips are on the plugs?

50 bucks for that cable is still rip off, you can write thousands of posts to me how expensive technology those wires have inside, I am still not convinced.

Hey, they sell even motherboards for about 50 dollars. There is little bit more I/O chips and stuff on a PC motherboard that in that your stinking LightPeak chip that in your opinion costs for Apple at least 10 dollars each?? LOL right... pffffff

oh crap it is 04:30am gotta get some sleep, lets argue some more tomorrow

----------

Quoted for truth. xinu appears to have no idea that pricing out a DSP that works with data at 10 gigabits is way different than at 100 megabits.

Say we decided to look up appropriate DSPs from TI to get a ballpark range.

Texas Instruments C5000 DSP is listed for $2-$10. (your guitar pedal)
Texas Instruments C6000 DSP is listed for $40-$200. (your thunderbolt port)

Throw in a couple of connectors at $17 per end and a pair of C6000s, and you're looking at a whopping $110 for a cable. At that kind of pricing, Apple's paying for your cable when you buy it at $50.

You can buy whole PC motherboard with 1gb networking, PCIe slots and all that, for around 50 dollars

Stop that nonsense about how costly some miniatyre I/O-DSP chip is. IT COSTS NOTHING TO MAKE !!!

 

[AidenShaw]

I wonder what people will bitch about once all the 2012 Macs arrive and have both Thunderbolt and SuperSpeed USB?

They'll be relieved that they can buy USB 3.0 drives, rather than paying 50% + $50 more for a T-Bolt drive. (50% more for the drive, plus $50 for the active cable)

But, you're right, they'll still be able to bitch about the mediocre graphics that Apple puts in the systems.

 

[hchung]

Why they didn't put those chips straight to Mac's and TB-devices? Why so stupid design that the I/O chips are on the plugs?

50 bucks for that cable is still rip off, you can write thousands of posts to me how expensive technology those wires have inside, I am still not convinced.

Hey, they sell even motherboards for about 50 dollars. There is little bit more I/O chips and stuff on a PC motherboard that in that your stinking LightPeak chip that in your opinion costs for Apple at least 10 dollars each?? LOL right... pffffff

You can buy whole PC motherboard with 1gb networking, PCIe slots and all that, for around 50 dollars

Stop that nonsense about how costly some miniatyre I/O-DSP chip is. IT COSTS NOTHING TO MAKE !!!

Apparently, you have no idea what kind of engineering effort goes into high frequency data lines. Without understanding that, "it's a wire with chips and plastic on the ends" is as far as you're going to be able to think about, and hence you're not convinced.

Seriously, there are people here who know way more about this stuff than you and I. And we're telling you $50 is cheap for this kind of cable. If you don't believe us and you're not willing to try to understand the physics behind it, then there's not much more we can do. It's like trying to explain to a guy who just discovered the wonders of overclocking that he will not be able to get his Core i5 to 10 Ghz.

And by the way, I speculate that the transceivers are on the wires because they need to be tuned individually to the specific wire's characteristics since not all wires behave the same. Somebody more fluent in this domain should correct me if I'm wrong.

Actually, here's something to do: Go and search for an Infiniband cable. Tell us what the average price is.

 

[Dr. Weezil]

The downside to fiber optic cables, however, is that devices connected using the longer fiber optic cables will require separate power cables, as running power over the cables at those distances is not currently practical.

This should be practicable not "practical."

Practical means "physical or tangible;" "practicable" means possible or feasible.

 

[KnightWRX]

What about the active chip? What's the price on this component?

http://www.ifixit.com/blog/2011/06/29/what-makes-the-thunderbolt-cable-lightning-fast/

Image

Probably quite a few bucks more than its worth. Some active chips go for as low as a few cents in quantities of a thousand.

Thunderbolt is made to be "overpriced" for consumers, because frankly, it doesn't seem aimed at the consumer market at all. USB 3 is both cheaper to implement and backwards compatible, which makes it quite more apt at consumer applications.

 

[repoman27]

Why they didn't put those chips straight to Mac's and TB-devices? Why so stupid design that the I/O chips are on the plugs?

50 bucks for that cable is still rip off, you can write thousands of posts to me how expensive technology those wires have inside, I am still not convinced.

Hey, they sell even motherboards for about 50 dollars. There is little bit more I/O chips and stuff on a PC motherboard that in that your stinking LightPeak chip that in your opinion costs for Apple at least 10 dollars each??

The Thunderbolt controller is in the Macs and the TB devices, and they cost $20-30. The chips in the cables are smaller and probably do cost only pennies apiece. There are however 4 Gennum chips and about 8 or so smaller ones in each cable, which probably add about $10 to the retail price of the cable.

That aside, $39 would still be pretty spendy for a 2m cable, and that's because these cables are actually challenging to manufacture. While there are passive copper twinaxial cables capable of handling 56 Gbps data rates, the Thunderbolt design is complicated by the additional requirements of supplying bus power and paths for out-of-band signaling. It is also a dual-channel, full-duplex design, which uses 4 separate pairs for high-speed signaling. It isn't easy to make a cable that can work at these speeds and also retain the qualities that are necessary for a consumer application, i.e. flexibility, minimum bend radius, strength, durability, heat dissipation, shielding. This is probably the first such cable to ever make it out of the datacenter or HPC environment and into people's homes.

You can buy whole PC motherboard with 1gb networking, PCIe slots and all that, for around 50 dollars

Stop that nonsense about how costly some miniatyre I/O-DSP chip is. IT COSTS NOTHING TO MAKE !!!

Show me the least expensive motherboard or PCIe add-in card that has 2 10GbE ports on it. The bottom line is that increasing performance by an order of magnitude is going to require you to wait a few years or shell out more money now. If you don't understand the difference between 1 Gbps and 2x10 Gbps, then Thunderbolt isn't for you.

Not sure I agree thats a good guage of adoption rates. Just because something is sold with it in doesn't mean its used. The number of peripherals would be a much better guage.

I've got a 2011 Air with TB but I certainly don't use it. But I do use USB3 in my hackintosh with my £10 USB3 cradle.

TB will stay an premium product just like FW except this time it has a much stronger competitor in USB3 in terms of fullfilling the requirement of a lot of users i.e. speed.

You're right, just because an OEM pushes something out into the market doesn't make it a good idea. I was mostly trying to illustrate that the number of Thunderbolt enabled motherboards shipped in the first year is very close to the number of USB 3.0 enabled boards to do so in the same amount of time. (Especially because the numbers I quoted for USB 3.0 were vastly inflated by the recall of motherboards due to the Intel 6-series SATA bug.) The USB-IF only certified 200 SuperSpeed devices in total in the first 14 months. However, this is still probably 5 times the number of Thunderbolt devices to become available in the same amount of time. Furthermore, sales of the USB 3.0 devices have been quite strong due to their relatively low cost and tangible benefit.

There are a few key issues that have hampered adoption on both sides. For Thunderbolt, Intel must supply Apple with one Thunderbolt controller for every Sandy Bridge processor they buy, and one for every ATD they produce. That has left everyone else waiting in line to get theirs. Once the supply pressures ease, we might see some progress. There's also the requirement to support DP++ on both ports of a 2-port Thunderbolt device. I'm not sure if the implementation that LaCie went with for the Thunderbolt LBD is based on a reference design, or if it's something they cooked up on their own, but it used a hell of a lot of silicon and board real estate to meet that requirement. Allegedly the soon to arrive Cactus Ridge controllers will integrate some of this functionality and make it easier to design and produce daisy-chainable devices.

On the USB 3.0 side, the problem was that those 200 certified devices were pretty much all the same. USB devices require bridge chips for each different device class, and initially there were only a couple available. If you wanted a storage device, rather bulky flash drive, card reader, or 4-port hub you were in luck, otherwise, no dice. Virtually all of the USB 3.0 hard drive enclosures on the market even today are limited to SATA 3 Gbit/s because there are very few SuperSpeed USB to SATA 6 Gbit/s bridge chips available.

Thunderbolt is not the best technology for simple external hard drives or flash drives, unless you own a 2011 Mac and it's your only option. In the long run, Thunderbolt has very strong potential for entirely new and innovative device classes. The people who bash it apparently can't see the uses for high-speed I/O beyond external storage enclosures. That being said, I will be very happy when Apple releases an OS update that has USB 3.0 drivers included so that I can use a $35 USB 3.0 ExpressCard in my MBP and connect a $120 2TB drive with no loss of performance. Or I could keep waiting until HDD prices drop back down to their pre-Thailand flooding levels and save another $30. :-/

 

[JazzyGB1]

This should be practicable not "practical."

Practical means "physical or tangible;" "practicable" means possible or feasible.

If you're going to be rude enough to pull people up on their use of English, you should at least ensure you're right.
His use of 'practical' in the context of the sentence was completely correct.

practical (prac|tical)

Pronunciation: /ˈpraktɪk(ə)l/
adjective

1: of or concerned with the actual doing or use of something rather than with theory and ideas:
Example: there are two obvious practical applications of the research

2: (of an idea, plan, or method) likely to succeed or be effective in real circumstances; feasible:
Example: neither of these strategies are practical for smaller businesses


You should learn English before taking others to task and more importantly, learn some manners!

 

[repoman27]

The MacBook Pro I bought recently has a TB input connector. What can be connected to it?

Here is a list of 20 currently shipping Thunderbolt devices and 1 cable that you can connect to your MBP today if you have the cash:

AJA

ioXT​

Apple

Apple Thunderbolt cable (2.0 m)
Apple Thunderbolt Display​

Blackmagic Design

Intensity Shuttle Thunderbolt
Intensity Extreme
UltraStudio 3D​

Elgato

Thunderbolt SSD (2 variants)​

LaCie

Little Big Disk Thunderbolt (3 variants)
2big Thunderbolt (2 variants)​

Matrox

MXO2 Thunderbolt adapter​

Promise

Pegasus R4 (2 variants)
Pegasus R6 (2 variants)
SANLink​

Seagate

GoFlex Thunderbolt adapter​

Sonnet

Echo ExpressCard/34 Thunderbolt Adapter​

 

[Rocketman]

New Prediction

I am now predicting the new Thunderbolt cable will have power and analog/digital sidegrades.

It will be something akin to or perhaps compatible with 1/8" audio.

It will have an optical TB center, and power and analog over exterior surfaces.

Rocketman

 

[hchung]

I am now predicting the new Thunderbolt cable will have power and analog/digital sidegrades.

It will be something akin to or perhaps compatible with 1/8" audio.

It will have an optical TB center, and power and analog over exterior surfaces.

Rocketman

I'd have to disagree.

It has to be backwards compatible with existing TB connectors. And based on what I read earlier from the dissections of the existing TB cables, I'd predict the following:

1) They'll have mini-DP connectors on the ends.
2) They'll have transceiver chips on both ends as well that convert the TB signal to optical (just like they do now for TB to copper)
3) Power will run parallel or coaxial to the optical.
4) Fully compatible with existing equipment. Capable of higher clock rates if the TB ports get upgraded.
5) Main advantage is longer possible length.
6) Every one of us including myself will immediately think "holy crap that's expensive for a cable" when it launches regardless of whether or not it really is overpriced compared to component costs.

 

[AidenShaw]

5) Main advantage is longer possible length.

Isn't that the *only* advantage?

If they'd drop the power over copper (optical-only) then a second advantage would be electrical isolation - but if there's copper that potential advantage is lost.

 

[Rocketman]

It has to be backwards compatible with existing TB connectors.

A dongle? They are replacing the 30 pin parallel connector with an optical one with power and analog sidegrade. Watch.

They are making far thinner devices soon.

X-serve in your pocket?

 

[hchung]

Isn't that the *only* advantage?

If they'd drop the power over copper (optical-only) then a second advantage would be electrical isolation - but if there's copper that potential advantage is lost.

For existing hardware, yes, that'd be the only advantage.
For future expansion of the protocol, no, optical makes it easier to transfer higher bit rates.

Both of these advantages stem from optical having less potential sources for interference. So if you wanted to narrow it down to a single distinct advantage, less interference would be it.

 

[AidenShaw]

For existing hardware, yes, that'd be the only advantage.

For future expansion of the protocol, no, optical makes it easier to transfer higher bit rates.

Both of these advantages stem from optical having less potential sources for interference. So if you wanted to narrow it down to a single distinct advantage, less interference would be it.

Of course, the "later this year" optical cables will probably never run faster than 10 Gbps - due to the transceiver and other active electronics built into each end of the cable.

Future revisions of optical cables would be needed for higher speeds.

I said "probably" because there's a miniscule chance that Intel could use higher speed optics and tranceivers - but that would probably add a lot to the cost of the cables.

 

[Lahmy88]

Wirelessly posted (Mozilla/5.0 (iPhone; CPU iPhone OS 5_1 like Mac OS X) AppleWebKit/534.46 (KHTML, like Gecko) Version/5.1 Mobile/9B176 Safari/7534.48.3)

Thunderbolt has a stupid name, your average user has no idea what it does.

I can see your point. 'USB' would have been instantly recognizable as 'Universal Serial Bus' to anyone and everyone when it first come out. Everyone would have known that it would be used for any and all computer periperals in the not-too-distant future. I'm of course being sarcastic. Why? Because the lesson is that all computer terms are not recognizable to your 'average user', only to us who give a crap (computer geeks).

 

[kustardking]

USB doesn't stand for Universal Serial Bus, it stands for "works"

I can see your point. 'USB' would have been instantly recognizable as 'Universal Serial Bus' to anyone and everyone when it first come out. Everyone would have known that it would be used for any and all computer periperals in the not-too-distant future. I'm of course being sarcastic. Why? Because the lesson is that all computer terms are not recognizable to your 'average user', only to us who give a crap (computer geeks).

Your statement should have ended with a period after "recognizable." Instead, you are showing how computer centric people such as *yourself* think. So, no, consumers would not recognize it as "Universal Serial Bus," but they *would* recognize it as familiar. What the acronym USB stands for is not important, rather the fact that XXX is on their phones, cameras, computers, keyboards, storage devices and speakers DOES matter. They can look at a box and find the three letters USB and know all is good.

 

[Ricky Smith]

Your statement should have ended with a period after "recognizable." Instead, you are showing how computer centric people such as *yourself* think. So, no, consumers would not recognize it as "Universal Serial Bus," but they *would* recognize it as familiar. What the acronym USB stands for is not important, rather the fact that XXX is on their phones, cameras, computers, keyboards, storage devices and speakers DOES matter. They can look at a box and find the three letters USB and know all is good.

Where's my UBS cable? That little square thing. The iPod port. That thing I connect my printer to.

80% of the average people I've seen don't even know what a USB port is.

 

[AidenShaw]

Where's my UBS cable? That little square thing. The iPod port. That thing I connect my printer to.

80% of the average people I've seen don't even know what a USB port is.

I wish that I had a UBS cable (Union Bank of Switzerland) to pump money into my bank account.

Your examples, however, actually go against your argument. The 80% know exactly what a USB port is ("the little square thing", "the Ipod port", "the printer connection") even if they don't use the TLA "USB" to refer to it. They know it's what they plug the keyboard, mouse, Ipod, phone, disk drive, thumb drive, whatever into.

And I think that the OP on the "USB naming" tangent wasn't referring to today, but to the idea that when PCs started to have USB ports (long before Apples had USB ports) the term "USB" didn't mean anything to almost everyone.

Like today, where most people have never heard of T-Bolt, except for the small niche of Apple users with new Apples who wonder if they'll get a terrible shock if they plug something into the hole with the lightning bolt glyph.

 

[Lahmy88]

so how long has USB been out for?
Ok, now how long has Thunderbolt been out for?
That's my point! It's barely been around long enough to be recognizable by the general public and I don't think there's an absolute intention for it to be a total USB replacement, more just a complementary technology (a much faster one at that )

 

[qoqo]

Whence Optical?

It's now 2013 and still no orderable TB fiber cables?

Someone PLEASE correct me if I'm mistaken (would love to be).

Still waiting to move the Mac out of the control room while leaving the Apogee DA/AD converters, and their 16 analog connectors in place...

 

[subsonix]

It's now 2013 and still no orderable TB fiber cables?

Someone PLEASE correct me if I'm mistaken (would love to be).

Netstor seem to offer them as an accessory, look at the very bottom of the page.

http://www.netstor.com.tw/_03/03_02.php?MTA4