Talk:Thunderbolt (interface)

From Wikipedia, the free encyclopedia
  (Redirected from Talk:Light Peak)
Jump to: navigation, search
WikiProject Computing (Rated C-class, Mid-importance)
WikiProject icon This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
C-Class article C  This article has been rated as C-Class on the project's quality scale.
 Mid  This article has been rated as Mid-importance on the project's importance scale.
WikiProject Apple Inc. / Macintosh  (Rated C-class, Mid-importance)
WikiProject icon This article is within the scope of WikiProject Apple Inc., a collaborative effort to improve the coverage of Apple, Macintosh, iOS and related topics on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
C-Class article C  This article has been rated as C-Class on the project's quality scale.
Checklist icon
 Mid  This article has been rated as Mid-importance on the project's importance scale.
Taskforce icon
This article is supported by Macintosh task force (marked as High-importance).

was the Origin of Light Peak project from USB 3 ?[edit]

Okay, bear with me here.. I remember that optical for peripheral interconnect was originally to be part of USB 3. So USB 3 was to have optical, where on the connector the center "bar" had fiber in it and it was still a passive cable. Then of course Intel split, because they couldn't get royalties for USB 3 controllers. Fair enough...

But the research done for this was renamed (or was always?) Light Peak which then we know that story.

Intel created USB. They opened up the USB standard back in the 1.0 days to get adoption in the industry. They were still a stong driver during 2.0. Then sometime when they were still involved with developing USB 3 they bailed out for the reasons above.

But of course now I can't find any evidence of this online these days. Maybe someone with greater Internet archive searching skill than I please find some evidence? Or am I totally crazy and mis-remembering this?

Also this wouldn't be to bias the article for pro/con USB, but the history is interesting and also explains some of the industry backlash against ThunderBolt. (talk) 03:05, 7 August 2013 (UTC)

i can't agree that "optical" entails any new changes and has anything to due with bus connectors or (USB/thunderbolt) version. (optical connectors on product: obviously differ)

for example: PCI or USB optical network cards use optics but their bus connection and also OS drivers are THE SAME. Only in the hardware bios would there be any difference, say for expected "window length" or dropped bits, or voltage/heat changes that needed handling differently. — Preceding unsigned comment added by (talk) 23:03, 16 July 2015 (UTC)

Rename to Thunderbolt?[edit]

Is not it totally the new name of Light Peak? [1]--I am a horny pussycat.Meow 14:38, 24 February 2011 (UTC)

Since Intel has now dubbed the technology as Thunderbolt, it seems right to rename the article to Thunderbolt and mention that it was developed under the name Light Peak. There a lot of double mentions at the moment, for example the first sentence. HectorMalot (talk) 16:29, 24 February 2011 (UTC)

There's a merge discussion for this at Talk:Thunderbolt (interface). --- Barek (talk) - 19:25, 24 February 2011 (UTC)

That merge discussion came out wrong. Soon we will have to deal with the confusion that Sony has launched "Light Peak" technology licensed from Intel, but which is not Thunderbolt – it uses the wrong connector, does not carry Displayport, et cetera. Intel marketing material, referenced in the beginning of the text, has caused this confusion. Also, the text is confusing as it stands – Thunderbolt uses a Displayport connector – demos by Intel uses USB connectors – so these demos per definition does not show "Thunderbolt", and so on. This is how these concept ought to have been used: Thunderbolt is a well defined concept – it should have an article. I am not sure Light Peak deserves an article – it is a collective name for technologies from Intel – and half a dozen partners - that can be used put together a protocol like for example Thunderbolt, and what Sony is doing with its Power Media Dock docking station. In this article, about Thunderbolt, we can refer to Light Peak technology when talking about Intel demos, and we can talk about Light peak as underlying technology on which Thunderbolt has been built. Power Media Dock will have it's own article which will refer to it also being built on Light Peak Technoliogy "like Thunderbolt". Oh, and copper vs fiber is not the differentiating factor beetween ”Sony Light Peak” and Thunderbolt – it is kind of a pretty safe bet that the future will hold a "Thunderbolt 2.0" that runs on fiber. I can rewrite this text but I will NOT waste time doing it if someone will just revert the changes, so what say you folks? gnirre (talk) 13:29, 28 June 2011 (UTC)

It is not totally clear to me what text you want to rewrite. I can see based on the information you provide, there were some wide misconceptions that Light Peak became Thunderbolt, when in actuality it is based on light peak technology. So in the hook, "originally code named light peak" is wrong and should say "based on light peak". Anywhere else in the article that suggest Thunderbolt was originally light peak should be changed as well, based on what you say. So I would certainly support your changes. It looks to me like at some point in the future, the original article on Light Peak may need to be resurrected since now not one but two connectors our based off it (and more in the future?). --TimL (talk) 15:18, 28 June 2011 (UTC)
I guess we are going to have to wait until there is a good source for this, though. I was hoping for Anandtech to sort this out, but they haven't gnirre (talk) 20:00, 28 June 2011 (UTC)
I think the best formulation is that Thunderbolt replaced Light Peak when Apple and Intel realized that an electrical interconnect was currently a more efficient solution than the interconnect using light that they had been working on. Sonys version of Light Peak is yet unnamed, as far as I know. Sony might not name it. It should be mentioned in Power Media Dock. I don't think Light Peak deserves its own article. There are most likely articles in wikipedia dealing with silicon photonics – they should have subheadings on Light Peak/Intel Silicon Photonics commerzialization or something like that. Light Peak could link there. gnirre (talk) 10:38, 29 June 2011 (UTC)
One more thing. We don't know if there ever will be a fiber Thunderbolt version. That is the plan, but plans change, as we have already seen. In light of that, it doesn't really make much sense to have a section called Introduction that exclusively is about Intel silicon photonics interconnect technology. It is a very interesting text, but it doesn't belong here. At least not until there exists a "Fiber Thunderbolt". It could be moved to Optical Interconnect or to Silicon_photonics. It did make sense when this article was titled Light Peak – but that's another discussion. gnirre (talk) 18:43, 29 June 2011 (UTC)
Given that the Light Peak and Thunderbolt articles now both redirect to the same page and that the only discussion here is about Apple's implementation of the technology, I've added in a mention of Sony using the optical version of the technology. Pdinc (talk) 03:40, 22 June 2012 (UTC)

No longer optical, but now copper based[edit]

It seems like Light Peak will be first introduced as a copper based technology according to various sources at CES 2011. I could not find concrete data but there are many articles out there talking about it. Here is one of them. The article should be updated accordingly when more info is available Frenchgizmo (talk) 00:49, 10 January 2011 (UTC)

Commenters believe in a limited bandwidth for copper, and the article should say that. But this too strong: "very much higher than what Copper Peak is expected to be capable of". I think that statement should be attributed (honestly, I am not convinced that it is true, so ... lets not put wikipedia in a state that it will need to retract from later). gnirre (talk) 08:39, 17 January 2011 (UTC)

on board[edit]

Apple seems to be on board, I say seems because so far it's only in mac blogs. Some of them must have references saying more than might/probably/secretly plan to. (talk) 10:35, 27 September 2009 (UTC)

This link may be interesting to you, then; it says Apple is the originator of the concept, and asked Intel to develop it. --moof (talk) 15:10, 27 September 2009 (UTC)

Powered cable[edit]

Will the Light Peak cable include electrical power to the gadgets? Thue | talk 19:21, 27 September 2009 (UTC)

Found the answer: Intel is working on it. Thue | talk 19:29, 27 September 2009 (UTC)
Measly 10 watts DC. Only a bit more than USB 3.0 (5) or Firewire (8), and nowhere near Power over Ethernet. This is a major advantage (along with standard cabling, cheaper switching, etc.) to Power over Ethernet over Thunderbolt for any powered applications. Daisy-chaining 10w across 7 devices? Really? This is effectively an unpowered interface that should be compared to USB 3.0 or unpowered 10 GbE, it isn't competitive with 1G powered ether for things like servo security cameras, etc., though it might be fine for small cameras and low-power VoIP phones. —Preceding unsigned comment added by (talk) 16:54, 25 February 2011 (UTC)
10 watts is a lot for peripheral devices. What are you wanting? 1.21 Gigawats? (talk) 14:01, 8 April 2011 (UTC)
"A lot"?!?? :Bill Gates said 640K was a lot of RAM in 1981. The article should list exactly what devices are and aren't feasible at the proposed 10 watts. And it should neutrally report that 10GB power over ethernet supports three times that much as a standard (30 watts) and much more (up to 80-90 watts) in its proprietary incarnations. In other words, 10GB PoE can run any existing monitor on the market, actually several of them, with one standard ethernet copper wire. And Thunderbolt can't and never will be able to, because those big screens will always take more than 10 watts. Most multimedia TVs sold as of 2011 had only a 100 megabit unpowered ethernet port in them anyway, and that was sufficient for the encoded stream output of a Blu-Ray (about 40mbps) or DVD (about 10mbps). Bandwidth matters less than the one-cable connection to a lot of people, and certainly matters more to integrators.
Confused comments from seemingly pro-Apple people should not have much status in these debates. The existing article is POV as it fails to really compare against the real standards being used in real fields now (VoIP, Wireless APs, security cameras are all power over ethernet now, and soon other industrial devices will have to follow).
/* Powered cable */ any comparison must honestly contrast the 0-watt Thunderbolt and 0-watt fibre 10GB Ethernet with the 9-watt USB 3.0, intended future 10-watt Thunderbolt, and (30, in practice up to 80-90) watt Power over Ethernet copper ether. Compare just to 1GB PoE for now since it's what's out in the marketplace. — Preceding unsigned comment added by (talk) 17:10, 12 January 2012 (UTC)

Competing optical technologies[edit]

Are there emerging/competing technologies with a similar goal and performance spec? Alanbrowne (talk) 20:51, 27 September 2009 (UTC)

On data transfer there is 10 Gbit Ethernet that is available today. However there is no solution to tunnel USB, SATA, HDMI/DVI or other protocols over it. There is AoE, iSCSI and NBD but little or no support in Firmware, so it cannot be used as a boot device. Tunnelling HDMI/DVI most likely also need hardware support, like a 10 Gbit Ethernet port on graphic cards. There are KVM solutions that can tunnel VGA/HDMI/DVI over Ethernet but they are not a full replacement for such a port, as they are intended for server management and at most handle gigabit Ethernet.

This is issues that should be further elaborated on. —Preceding unsigned comment added by (talk) 09:17, 25 November 2010 (UTC)

Transfer Rate[edit]

Gb/s = Gigabits per second, not Gigabytes. Can someone please clarify which of these two is the correct transfer rate? —Preceding unsigned comment added by (talk) 08:13, 10 October 2009 (UTC)

It is 10 Gbps (10 Gigabits per second) per channel. There are 2 bi-directional channels. See Dan.oliver (talk) 20:58, 9 March 2011 (UTC)

Suggested characteristics by public section[edit]

I'm just curious where this section came from. I've performed several google searches on the subject, and haven't come up with anything. Who are the "public" this section is referring to? Cyclonius (talk) 03:00, 25 December 2009 (UTC)

Any new info?[edit]

It's the middle of Februrary 2010... and there been any new developments with Light Peak? -- (talk) 16:54, 23 February 2010 (UTC)

Bulky cables[edit]

Optic cables aren't bulky in general. Only multi-fiber types designed for "outside" use are equipped with heavy armouring. Those used to connect devices in racks are even thinner than an ordinary USB cable. I admit, however, that more convenient plugs should be developed for customer use (OFC_ST) but unlike USB and FW they still should feature latches. Stlman (talk) 19:37, 20 March 2010 (UTC)

Afaict the patch cables used inside racks are indeed thin BUT afaict they are designed to be installed once and left alone within the relative safety of a rack. Light peak cables OTOH will need to be ruggedised sufficiantly to handle consumer use (e.g. it should be very difficult to damage one accidently by over-bending it). Plugwash (talk) 15:07, 4 December 2010 (UTC)

ClearCurve speed[edit]

Even the single-mode version, with a single carrier frequency, offers maximum data rates of 25 Gbps.[4]

If I remember correctly my physics courses, monomode fibers are always the fastest, as you don't have any modal dispersion; only chromatic dispersion (insignificant). So the word Even is superfluous or wrong.

I'm not totally sure (that's why I post this comment and don't edit the page) but I guess it should be replaced by

The monomode version offers maximum data rates of 25Gbps. [4] --7e'o (talk) 08:40, 24 March 2010 (UTC)

Connector picture[edit]

Is there a picture of any connectors for cables using this standard? Also, is there already a pictorial convention on how the connectors would be labeled? Demf (talk) 18:12, 15 April 2010 (UTC)

To this date the connectors are stil not defined. For the prototype shown during IDF 2010 Intel used a modified USB connector. (talk) 11:27, 20 September 2010 (UTC)

chances are it will use the prototype connector, due to it now being extensively used for over 3 years in testing by intel&partners. Markthemac (talk) 21:41, 29 November 2010 (UTC)
The connector pin out is documented in Apple's patent specification. See here. According to the latest documents that Intel has recently revealed, this is in fact the final pin out as it is in use now for quite a while. (talk) 17:19, 11 April 2012 (UTC)

Large section removed[edit]

In April a single-edit anon editor removed a large portion of the article with no notice, checkin note, or anything of the sort. Should this be reverted? Maury Markowitz (talk) 14:41, 28 July 2010 (UTC)

I have done so, let's see if anyone pops up to explain the deletion. (talk) 18:04, 31 July 2010 (UTC)

Removing "written like an advertisement."[edit]

I'm of the opinion that it's going to be difficult to write a truly NPOV article about technology that hasn't been released to the public. This is to be expected because the company has a near monopoly on all the information regarding their product.

That being said, I'm going to go ahead and remove the "written like an advertisement" header. I think the article does a pretty good job of maintaining an encyclopedic tone and, when I read through it, I didn't notice anything that sounded like advertising.

If you believe that the WP:SOAP tag still belongs, please provide explicit examples of where the article needs improvement. If something is not properly cited, well, that's what [citation needed] is for. Robert Seaton (talk) 16:36, 14 September 2010 (UTC)

Reads with USB bias currently. Phil.andy.graves (talk) 12:36, 15 October 2010 (UTC)
No "USB bias" evident now. The article seems well balanced now that the fact that intel has admitted that it can't deliver fibre at all as of January 2011, and has announced strong and unequivocal support for HDMI and DisplayPort, and of course will also be shipping 10 Gb and 100Gb Ethernet and USB 3 devices. The article seems to skeptically analyze the reasons why a single fibre optic cabling standard will not displace these purpose-specific standards (USB for peripherals, Ethernet for networking, DisplayPort for monitors) before at least 2014 or so. Those reasons being:
No clarity on the power interface. Now that there's a copper interface, will it include power? No answers from Intel, and it's pretty late in the game not to have that clarified.
Need for certainty among makers of consumer items like televisions and lower end commercial items like monitors. These tend to outlast desktop and laptop systems and stay in use for ten years or more (how many 10 year old monitors do you have in your house? what are they doing? when you replace your box do you replace your monitor too, or only with every other box or so?)
Inability to deliver fibre - Copper Peak basically admits that whatever advantages the process of IC fabrication for Light Peak has, it just cannot make the interface and connector and cable so cheap as to displace copper just yet, and probably not for five years.
No compelling reason to believe that 10 Gb, 40 Gb and 100 Gb Ethernet will not come down in price (as 1 Gb Ethernet did and 100 Mb Ethernet before it) and be used for peripherals, monitors and other high-bandwidth devices. Already we see a lot of media drives and devices supporting Ethernet, and that includes HDMI (HDMI 1.4e). Blu-Ray only needs about 40 Mbps to move its encoded data stream off the disc, so HDMI 1.4e's 100 mbps interface seems good enough until Hollywood re-encodes everything. Is that really going to happen any time before 2020? When they do, is the next interface really going to require even 100 mbps, or will encoding be so good they can get 4x the resolution through only 2.4x the bandwidth? HDI 1.4e is a strong bet that 100 mbps is enough. —Preceding unsigned comment added by (talk) 03:41, 12 January 2011 (UTC)


When does this techonology actually come to the marked? (probably a stupid question, but i didn't figure it out while reading the article). When will Intel push for the adoption of the technology?

Sorry for reposting from other articles, but i'll put it in here anyway: In December 2010 it was announced that Intel and AMD, with backing from various computer vendors, would stop supporting DVI-I, DVI-A, VGA and LVDS-technologies (i.e., Firewire) from 2013/2015, and instead speed up the adoption of DisplayPort and/or - VGA Given 5 Years to Live (9. December 2010).

"Legacy interfaces such as VGA, DVI and LVDS have not kept pace, and newer standards such as DisplayPort and HDMI clearly provide the best connectivity options moving forward. In our opinion, DisplayPort 1.2 is the future interface for PC monitors, along with HDMI 1.4a for TV connectivity." - AMD, Dell, Intel Corporation, Lenovo, Samsung Electronics and LG. Dec 8, 2010. [2].

Since they apparently seem happy with DisplayPort and HDMI and say it's the future, are they allready now planning to campaign against it?

From the first source: "IDC's figures show DisplayPort was on 5.1 percent of commercial desktops in 2009, but that figure will grow to 89.5 percent of them in 2014. In commercial notebooks, DisplayPort's penetration will increase from 2.1 percent in 2009 to 95 percent in 2014.

Only 24.5 million of the 427 million laptops in users' hands in 2014 will be VGA-enabled, Daoud stated. Another 279 million will use HDMI, while 167 million will use DisplayPort."

Does Intel work for the adoption of DisplayPort and HDMI now, and then want them removed later? Sometime after 2015 perhaps? Or will this hit the marked next year? :-S

The information sources are at least coherent about firewire: "Both Intel and AMD will also stop supporting low-voltage differential signaling (LVDS) by 2013." was stated now December 8. Yosh3000 (talk) 23:40, 27 December 2010 (UTC)

10Gb Ethernet section: reach, comparison, power capabilities[edit]

This section states that for 10Gb 'Ethernet cable is supported but using copper cables the maximum reach is only 5 to 7 meters'.

10Gb Ethernet is supported up to 100m on Cat6A copper twisted pair. (See 10G-BASE-T).

Biff77 (talk) 19:44, 20 January 2011 (UTC)

I have made the corrections you suggested, but in the future, please consider making them yourself! I'm not sure if the section should be relabelled as I am unsure if 10 Gb Ethernet over twisted-pair copper is still unpowered. -- (talk) 15:06, 23 February 2011 (UTC)
10 Gb powered standard isn't out yet, is it?
On the reach: the majority of 10GbE equipment only practically supports a short run. The 100m distance is expensive to support. So a different wording like "only 5 to 7 meters on most commercially available equipment as of 2010" would be more accurate than citing the 100m number.
Now that the actual interface is out there, a more direct comparison table should maybe be put together. Notice that daisy-chaining 10w across 7 devices means basically the interface is effectively unpowered - 10 watts divided even by two is not enough for most 10G speed devices. So in practice when using the power the daisy-chaining is of no value, and when not using it the proper comparison is to 10GbE. This should be reflected in the comparison charts now needed.

Merge with Light Peak[edit]

Done. /ƒETCHCOMMS/ 04:14, 25 February 2011 (UTC)
The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Since Light Peak is just a codename and Thunderbolt is the final name as according to Intel, the old article should be merged to this one, what do you think? -- (talk) 15:12, 24 February 2011 (UTC)

  • AGREE riffic (talk) 15:16, 24 February 2011 (UTC)
  • Yes (talk) 15:25, 24 February 2011 (UTC)
  • Agree. Acabashi (talk) 15:26, 24 February 2011 (UTC)
  • makes perfect sense —Preceding unsigned comment added by (talk) 15:28, 24 February 2011 (UTC)
  • But, but, but, think of all the poor minority who will forever call it Light Peak!  ;) YES, completely agree with it (LP should have been simply renamed IMHO). Light Peak, sounds like bright flash (Peak of Light), so Thunderbolt sounds very appropriate. I hope someone will reveal how name was chosen.  :) --Flightsoffancy (talk) 15:34, 24 February 2011 (UTC)
  • Yes, do it. Tweisbach (talk) 15:38, 24 February 2011 (UTC)
  • I fully agree. As Intel has renamed the technology (now marketed to the public) as Thunderbolt, I think people are going to look there for info.--Paracel63 (talk) 17:21, 24 February 2011 (UTC)
  • Disagree. I think that we should wait for more information on Thunderbolt before any merging is done. As far as I know, the implementation in the new MacBooks doesn't actually have any optical lines, because Intel was able to do 10Mbps using copper. So my suggestions are:
    1. If you want to merge anything, then merge Thunderbolt into LightPeak as it's early implementation. You can always rename LightPeak to Thunderbolt later.
    2. Or don't merge it at all and wait. Just add links to both the articles so users can quickly get to the other one. --User:Nolwendil —Preceding undated comment added 15:46, 24 February 2011 (UTC).
    Quoted from Intel: Developed by Intel (under the code name Light Peak), . Their is no debate, Light Peak is now called Thunderbolt by Intel, period. --Flightsoffancy (talk) 16:50, 24 February 2011 (UTC)
    OK then. I agree it should be merged. However, it should be stated that it is not yet certain, whereas the early implementation on the MacBooks contains any optical connections, or if it is metalic-only. --User:Nolwendil
  • Hmmm, I don't agree. Light Peak should have its own page :D —Preceding unsigned comment added by (talk) 16:22, 24 February 2011 (UTC)
  • Disagree. Thunderbolt is Apple's marketing name for Intel's technology. It is very much in doubt if other PC manufacturers will use the name Thunderbolt. This is very much like the IEEE934 article and the terms Firewire (Apple), iLink (Sony) and Lynx (TI). At the most, Thunderbolt could possibly have a section within the Light Peak article if Apple's implementation differs from "the standard". — Preceding unsigned comment added by Mconwell (talkcontribs) 16:59, 24 February 2011 (UTC)
    The last post is wrong. Intel has named it Thunderbolt. (talk) 17:03, 24 February 2011 (UTC)
  • Merge ʍαμ$ʏ5043 (talk) 17:21, 24 February 2011 (UTC)
  • Merge. Ttemiel (talk) 18:04, 24 February 2011 (UTC)
  • Agree. Thunderbolt is the official name, LightPeak is to Thunderbolt as Longhorn was to Vista. Let's change the article name already. Althepal (talk) 19:23, 24 February 2011 (UTC)
  • Agree. This shouldn't even be a discussion. Intel say Thunderbolt is the official name, Apple are using it too. "Light Peak" was the codename, just mention this in the article. --Mwongozi (talk) 19:31, 24 February 2011 (UTC)
  • Agree, but the 'Light Peak' page should be renamed and the information in the existing 'Thunderbolt' page be merged into it. --Spazturtle (talk) 20:57, 24 February 2011 (UTC)
  • Agree. Merge. Intel says Thunderbolt's the official name now.PRRfan (talk) 21:19, 24 February 2011 (UTC)
  • Agree, but the Light Peak page should be renamed to Thunderbolt and the current Thunderbolt page should be merged into the page. That way we preserve the history of the Light Peak page from the last year. –TheIguana (talk) 22:41, 24 February 2011 (UTC)
  • Agree Merge away! There is no point in two articles since the tech is one and the same. Definitely keep the history though! —Preceding unsigned comment added by (talk) 01:12, 25 February 2011 (UTC)
  • Support merging. This will also require a page-history merge, I think.  --Lambiam 22:45, 24 February 2011 (UTC)
Note: I would do the history merge myself, but I'm not using my admin account from this PC. I've submitted a request at WP:ANI#Merge needed at Thunderbolt (interface) to have another admin assist. --- Barek (talk) - 22:55, 24 February 2011 (UTC)
  • Then can an admin merge it? I am the original writer of this debate, I am sorry that I am no longer registered on Wikipedia, the IP I am on is home, not like the one I was using earlier, anyways, it seems that everyone agrees that it should be merged. Can some one do that? The easiest seems to copy all the info from Light Peak over here and redirect that page to this one. -- (talk) 00:13, 25 February 2011 (UTC)
I posted above, I've requested a history merge at WP:ANI, so that the change history from Light Peak can be moved over here (as that one is the more developed version of the article, and has the longer history log for attribution). --- Barek (talk) - 00:17, 25 February 2011 (UTC)
That's a blessing and a curse, because it's a verbose article full of incorrect speculation dating back years. Aeiuthhiet (talk) 00:35, 25 February 2011 (UTC)
  • Agree. Light Peak was Thunderbolt's code name during development. Oaakroll (talk) 02:34, 25 February 2011 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.


Per the above snow consensus, I've histmerged. Light Peak is the live revision, original Thunderbolt (interface) is in this revision; please pull relevant content out and add it to the live page. /ƒETCHCOMMS/ 04:14, 25 February 2011 (UTC)

In hindsight, this was probably a bad idea. --TimL (talk) 02:28, 8 July 2011 (UTC)

Include image of the present interface?[edit]

is the image from the Thunderbolt promotional page linked in the "factual accuracy: out of date" notice.
Is this something that should be included in the article at this time?
Sire TRM (talk) 04:38, 25 February 2011 (UTC)

Protocol independency[edit]

What about the original plans that Light Peak would be protocol independent, and SCSI etc. could be run over it? IIRC it was mentioned in the original Light Peak article last year. SyP (talk) 07:42, 25 February 2011 (UTC)

You can tunnel PCIe over it. I think that should suffice for driving a SCSI controller, latency issues notwithstanding. There are probably other ways you could tunnel it as well. Dgatwood (talk) 23:39, 14 April 2011 (UTC)

Products that use Thunderbolt[edit]

Since Thunderbolt is so new, I think it would be good to have such a section (until the products become widely available). The only ones I'm aware of is the Macbook Pro (available) and the Lacie Little Big Disk (announced). Drrll (talk) 11:29, 25 February 2011 (UTC) Also used on Pegasus RAID storage with Thunderbolt. John a s (talk) 17:21, 24 July 2011 (UTC)

Agree. Someone should add a new section discussing the varies announced and shipping Thunderbolt products. There aren't too many haha. Photographerguy (talk) 07:10, 15 March 2012 (UTC)

I added this section to the end and moved Apple computers from market introduction to there. Also added a few products there. Didn't have time to add references but they will among top google hits. Market introduction chapter might need some minor adjustments after this to have a logical structure. Golemus —Preceding undated comment added 00:30, 28 April 2012 (UTC).

Apparently the product list was removed by Oosh. I really don't understand what is the problem with listing currently available products. Of course the list will (hopefully) grow rapidly so yes maybe it is better in its own article and even in that hopefully in 1-2 years there will be so many products that there is no use of a list. But currently it is of value for many people as you see. I haven't used Wiki for a while so maybe somebody more capable could make the new list article and put the products there from the revisions I did about a week ago. — Preceding unsigned comment added by Golemus (talkcontribs) 12:35, 3 May 2012 (UTC)

Wikipedia is not a directory (WP:NOTDIRECTORY), as I put in the changelog we don't maintain lists of devices that support USB, IEEE 1334, DisplayPort, etc. If it is notable (WP:NOTABILITY) that a device supports it, for example a first, then by all means mention it. But having an ever growing list of devices is not the domain of an Encyclopaedia. At best I can see merit in a the creation of a new category i.e. 'Devices that support ThunderBolt' and tagging device articles appropriately. -Oosh (talk) 13:16, 3 May 2012 (UTC)

I added 'list of thunderbolt compatible products' and took the product list to there from and old revision. There is some citation error and I am not very familiar with wikipedia to correct it, somebody please.. :), and of course add all the products that you know. The list will become obsolete in 1-2 years but I believe that now it is of high value for many people as most of the new laptops and motherboards do not have thunderbolt currently. — Preceding unsigned comment added by Golemus (talkcontribs) 16:36, 22 May 2012 (UTC)

Wikipedia doesn't need to list every single Thunderbolt device ever. It should, however, list the first devices available, the class of devices available, the most recent devices, the major companies offering devices and which markets they are going into, etc, etc. And while there are few Thunderbolt devices on the market, it could well mean listing all of them. Currently the article lists approximately nothing, which is not very helpful for someone trying to learn what Thunderbolt is and how it's used in the real world. —Pengo 21:55, 22 August 2012 (UTC)

Still written like an advertisement[edit]

As the new update of the MacBook pro nears its release, many more people are going to be drawn to this page to find out what Thunderbolt is. At this point, the article still lacks neutrality - most sections either describe details of the interface in a positive light or compare it negatively with USB 3.0 and firewire. The article mentions that you can daisy chain seven devices but does not compare that with the superior chaining capacity of USB 3.0 and firewire. What do you think? Dsi2104 (talk) 17:19, 25 February 2011 (UTC)

Many changes were made today that have extensively improved both the quality and the neutrality of the article as well as providing missing citations. I am moving the NPOV tag. TimL (talk) 04:16, 26 February 2011 (UTC)
The problem is that nobody else is making any devices (chipsets), if this standard catches-on things will change. Markthemac (talk) 22:57, 26 February 2011 (UTC)
Two statements that apply is Wikipedia:NOTADVERTISING and Wikipedia:Neutral point of view. If this article compares other standards such as USB and Firewire, then it should discuss both PRO's and CON's of such standards, not just the CON's. • SbmeirowTalk • 00:41, 28 February 2011 (UTC)
The section on USB 3.0 reads more like an advertorial of advantages of Thunderbolt over USB 3.0. In particular the references to compatibility of USB 3.0 and 2.0 are irrelevant. — Preceding unsigned comment added by (talk) 00:11, 7 July 2011 (UTC)
The USB 3.0 section has been removed between July 7 and today. Only fleeting mentions of USB explaining that Intel originally sought/contemplated to use USB as the physical form factor remain. Anything else looking remotely like an advertisement? Bmike8 (talk) 20:46, 17 July 2011 (UTC)


According to this source, Thunderbolt may be just as 'risky' as Firewire, since it may provide direct access to memory through the PCIe bus. (talk) 20:55, 25 February 2011 (UTC)

that's very speculative Markthemac (talk) 22:58, 26 February 2011 (UTC)

Continuous Transfer Rates[edit]

I'm stating the following to ensure authors don't forget to take this into account when comparing bus speeds.

Everyone that praises faster bus speeds always avoids talking about real-world continous transfer rates, because at the end of the day the slowest things in a system will determine the maximum transfer rate. This is a technical fact that is avoided in all marketing hype.

A combination of many things on both the host and device side will determine the maximum transfer rate: hard drive rotation, hard drive head seek times, memory bus transfer rates, software overhead, interface transport time, and more.

The following is an example of large delays that all faster buses will never be able to speed up.

Seagate Barracuda 7200.11
  • 7200 RPM, which is 0.128 mSec for one revolution.
  • Random read seek time <8.5 mSec.
  • Random write seek time <9.5 mSec.

Hard drive delays destroy almost all speed advantages of a higher bus speed, which is likely why they decided to use SSD (flash-based) media for the very first Thunderbolt external drive.

Even a fast external SSD drive will still be rate limited by the hard drive in the host computer when copying files between drives, but hard drives on both sides will show the true limitation of faster buses. • SbmeirowTalk • 18:04, 28 February 2011 (UTC)

You are referring to limitations of peripherals, not limitations of buses. "at the end of the day the slowest things in a system will determine the maximum transfer rate." False. This depends on the capabilities of the bus. A bus can potentially handle multiple peripherals up to it's maximum speed minus protocol overhead, which varies from bus to bus. I don't think even the marketing asserts that buses make peripherals faster. TimL (talk) 02:06, 4 March 2011 (UTC)
Limitations of storage devices is my exact point, which I wanted to make sure that authors on this article consider their wording to ensure it doesn't sound like a external Thunderbolt drive is NN times faster. I see this happen all the time with the technically incompetent writers of news articles and blogs, and I want to ensure the wording in this article doesn't accidentally say the wrong thing. Since numerous people are involved in writing this article, I now have confidence that it will be correct. • SbmeirowTalk • 07:51, 4 March 2011 (UTC)
The statement "Hard drive speed could be the cause of Thunderbolt not getting closer to its theoretical maximum." does concern me, and I think it should be slightly reworded. • SbmeirowTalk • 07:54, 4 March 2011 (UTC)


Thunderbolt does not provide 8 ns latency. 8 ns is insufficient to even travel the 3 meter cable length at the speed of light. 8 ns refers to the time synchronization provided by Thunderbolt.

ArbitraryConstant (talk) —Preceding undated comment added 23:47, 2 March 2011 (UTC).

Copper vs. optical section: optical cable type?[edit]

The Copper vs. optical section indicates that the optical cables being developed "...are to have two 62.5-micron-wide fibers to transport an infrared signal up to 100 metres (330 ft)." The video at is cited, but there is only a mention that "this fiber is 125 microns wide," which is the size of the cladding diameter. At no point is the core diameter specified to be either 50 or 62.5 micron.

The maximum distance of 100 meters seems to indicate that the fiber cable would use laser-optimized 50/125 micron multi-mode fibers (OM3 or better), but this would need to be verified. It may be best to change the section to indicate that the cables "are to have two 125-micron-wide fibers to transport an infrared signal up to 100 metres (330 ft)" and leave it at that until more details are forthcoming from Intel. Wells.brian (talk) 03:07, 7 March 2011 (UTC)

Peer to peer?[edit]

A huge limitation of USB 1.0/2.0 was the distinction between devices in host mode/client mode. If my camera was made to connect to a PC, it could only talk to a PC and not connect to a USB hard drive for more photo storage. Firewire was a peer-to-peer setup where any two connected devices could communicate. Does Thunderbolt work this way? — Preceding unsigned comment added by Donpdonp (talkcontribs) 15:52, 17 March 2011 (UTC)

No, Apple's documentation sez it's just like USB. Much like USB though, clunky extra hardware can be used to make host devices with such stuff masquerade as clients. (talk) 14:35, 22 October 2012 (UTC)


What is this doing here?

"Unlike USB, to redirect PCI to other (virtual) machines, device specific solutions are required."

It's wrong, for one thing; there's no reason in principle that you couldn't provide a PCI device to a virtual machine as long as the host OS isn't doing anything with it. You can't usefully do a true interrupt handler for PCIe (you have to use software interrupts), but then again, there's no such thing as a hardware interrupt for USB, making this rather moot.

Second, Thunderbolt is largely used by tunneling PCIe data, which means that anything that applies to PCI devices also usually applies to TB devices. Why is the article comparing USB to PCI in an article about Thunderbolt?

Is there something I'm missing?

Dgatwood (talk) 23:47, 14 April 2011 (UTC)

USB 3.0 Bias[edit]

The section containing comparisons to different transfer technologies comes off as very biased. How is the line in the USB 3.0 section appropriate at all? "The USB protocol has historically had a variety of problems when operating at high speeds" Totally unnecessary. IMHO, the whole article needs to be stripped down to just the specs and take out all the opinions and analysis. The acceptance of Thunderbolt vs. USB 3.0 as the new widespread transfer standard means big things for Intel, Apple and every other computer company. —Preceding unsigned comment added by (talk) 05:11, 16 May 2011 (UTC)

Removed comparisons section[edit]

Based on the feedback here, this was confusing to readers, in addition much of it lacked citations. It's too new a technology, as is USB 3.0, to make encyclopedic comparisons yet. --TimL (talk) 02:37, 8 July 2011 (UTC)

I wasn't part of those giving negative feedback, but an additional fault which appeared within that section is that "while previous USB implementations typically reach less than half of their theoretical capability" is [not in citation given] and simply untrue:
  • the citation didn't state anything about "half"
  • nor did the cited article lump USB3.0 together with the OTHER "previous USB implementations" -- and for good reason: it does a disservice to readers because USB3.0 should be lumped in with LightPeak, not USB2.0-1.0, when it comes to real-world-versus-theoretical speeds: USB3.0 typically reaches 2/3rds of its theoretical capability, which puts USB3.0 in the same league with LightPeak, and USB2.0-1.0 give vastly different/inferior real-world-versus-theoretical speed caps compared to LightPeak/USB3.0.
Additionally, micro-USB3.0 was released awhile ago, and matches (or actually is smaller than) the 5mmx8mm size of Thunderbolt, in contradiction to: "The higher speeds required changes to the cabling, with all plugs now substantially larger than Thunderbolt. {no citation}"
Micro-USB3 is 12mm wide. I don't think we need a citation to say that 12mm is bigger then 8mm. (12mm is my measurement, I get conflicting info from google search.) Algr (talk) 05:49, 13 April 2012 (UTC)
. . . and "correct wiring can be difficult for casual computer users" is a statement of opinion, uncited.
This is all in addition to the critiques made in the last 2 topics immediately above, and the "Still written like an advertisement" topic above those. (talk) 11:33, 8 July 2011 (UTC)

Needs new comparison section![edit]

Removing that comparison section doesn't mean there should be no comparison section. Previous versions of this article did a very good job of comparing against other technologies, though they should probably have been in table format. Items to be compared include: Power capability and wattage (zero in Thunderbolt now, zero in any future fibre implementation, 4.5 watts for USB 3.0, 30 watts for Power over Ethernet copper, zero for 10GB or 100GB fibre ethernet, cost per port (maybe several years' worth to show trends)). Agreed that Firewire and USB 2.0 simply don't belong on the comparison list, these are obsolete technologies and no one buying a motherboard or laptop today is going to be comparing either to Thunderbolt.
As soon as there is a motherboard-add-on combination for generic intel/AMD motherboards, maybe from ASUS, some non-MacOS-based comparisons can be easily made and must be in this article. Meanwhile there are some useful statements in prior versions, especially regarding the power situation and the real purpose of Thunderbolt (like Firewire) being to drive down 10GB power over ethernet technology to be affordable (just as Firewire drove down 1GB ethernet to be affordable). By no means should the article give the impression that Thunderbolt has a future beyond that. There is no proof of its scalability (unlike Ethernet), cable length is poor and it has no power connector. Why would anyone use it if they could get a 10GB PoE connector for the same price per port? There are no devices capable of outrunning a 10Ggbps port anyway as of early 2012. — Preceding unsigned comment added by (talk) 17:03, 12 January 2012 (UTC)
What do you mean by "Zero in thunderbolt now"? Apple's site says that all the ports provide 10 watts, and several bus powered thunderbolt drives exist. When was there ever an unpowered thunderbolt port? Algr (talk) 05:24, 13 April 2012 (UTC)

PCIe Doesn't Carry Video Data[edit]

"Because the PCIe bus does not carry video data, it is unclear whether a standalone PCIe card could offer a Thunderbolt port."

This sentence is completely unclear. The PCIe does carry video "data", but not in a very useful form. The sentence needs to to be clarified with exactly what the PCIe does and doesn't carry relative to video. Jlodman (talk) 19:20, 19 January 2012 (UTC)

While it would hypothetically be possible to tunnel PCIe through TB and connect to an in-monitor external GPU, this is pretty unlikely. The only present TB monitor (Apple's) uses DP data tunneled through TB, so a hypothetical TB interface without DP inputs wouldn't work with it. On that note, I can also imagine a TB card with DP passthrough for existing GPUs, sort of like the old 3DFX VooDoo GPUs. (talk) 14:35, 22 October 2012 (UTC)

public available Specification[edit]

Exist any public available Specification for Thunderbolt? I am interested for the physical Layer of the Thunderbolt-Interface. Are the 10 GB/s with or without Coding? 8/10 or 64/66 or 128/130 Coding? Is the Cable a Cross-Over? A stuff in a shop says the direction of the cable is arbitrary. Thanks for any helpfully reply. Erik (talk) 21:41, 30 April 2012 (UTC)

Intel is hiding all the good stuff behind paywalls, not even so much as a whitepaper has leaked out. The best I could find is an SDK guide from Apple I added to the article, which contains a little hardware details (mostly pretty specific to the Light Ridge controller they're using now.) To answer your questions, it's 10GBps with 64/66 coding (though it appears PCIe/DP lanes are only allowed whole through TB lanes, so the lack of granularity will limit throughput,) and it is indeed crossover. (talk) 14:35, 22 October 2012 (UTC)
Thank You! Erik-- (talk) 21:56, 25 February 2013 (UTC)

What is Thunderbolt (in English)[edit]

I read the first three paragraphs of this article and I am no nearer knowing what it is about. I gave up after that. Cannonmc (talk) 23:47, 27 December 2012 (UTC)

That's a bit of an exaggeration, to say the least, but I agree the lead of the article grew too large. I've rewritten it, kicked the description section above the history one, and merged the balance of the lead into the description. (talk) 15:59, 2 January 2013 (UTC)

No, it's not an exaggeration, it is the truth - I didn't understand it. I had no idea what Thunderbolt was. I still don't. I will re-read your changes but I wish Wikipedia authors would realise that they are (or should be) talking to non-specialist readers. If I understood what the article was saying I wouldn't need to read it in the first place.

It wasn't whether the 'lead' was too large, it was whether it was written in plain language that could be understood by a 'layman'

Cannonmc (talk) 17:37, 5 January 2013 (UTC)

"Thunderbolt combines PCI Express (PCIe) and DisplayPort (DP) into one serial signal alongside a DC connection for electric power, transmitted over one cable"

That means nothing to the uninitiated unless they follow another (at least) two links. Explain in plain language...please. What is "PCI Express (PCIe) and DisplayPort (DP)" without using technical jargon? Cannonmc (talk) 17:42, 5 January 2013 (UTC)

Other interfaces for connecting stuff, but that's not necessary to know. If you don't own or aren't interested in products intended for use with PCIe or DP connectors, all you're probably interested in is that Thunderbolt's a “…hardware interface that allows for the connection of external peripherals to a computer…” as noted in the first paragraph. Aside from that, maybe that it's a very fast interface.
The second paragraph is simply to tell anyone familiar with PCIe or DP what Thunderbolt is, and to provide convenient links to the articles for them to those who are curious. Any description of PCIe or DP beyond “they're peripheral interfaces too” would be redundant to their articles (DP specialized nature being self explanatory, it's a port for displays.) (talk) 15:04, 22 January 2013 (UTC)
Agree with the point made by Cannonmc. Perhaps an example of its usage would be better than saying what it combines. Is it for connecting external storage devices, display equipment or both? At the moment I can't decide if it's a way to attach half-a-dozen monitors or a few external backup drives :) — Preceding unsigned comment added by (talk) 14:32, 19 March 2013 (UTC)
if I understand right its intended use is what you put in the pci express slot on the motherboard in the computer, but external. But since the market for external graphics cards is quite limited, it is instead used for external harddisks connected to Apple computers with extra expensive active cables, and maybe docking stations? — Preceding unsigned comment added by (talk) 22:08, 28 August 2013 (UTC)
Its intended use is a high-speed connection for high-performance peripherals. Imagine USB3 but much faster and more diverse. You can, for instance, connect 3 4K displays via Thunderbolt, or connect a single Thunderbolt cable to a display, and use that connection to also carry a USB hub, webcam, and other peripherals either built into the display, or daisy-chained to it. It uses principles already in place (specifically PCI) to make this an easy integration, rather than being all-out new. I'll see what I can do to make this clearer in the article.  drewmunn  talk  08:34, 29 August 2013 (UTC)
Hi, 12 months on, I am afraid I have to reiterate the earlier comment: I read the opening paragraphs and I still don't know what it is and what it does. Perhaps things like "high-performance peripherals" could be unpacked. Slac speak up! 06:01, 8 December 2013 (UTC)

See Fulgurite -- (talk) 22:16, 23 February 2015 (UTC)

cite web publisher reference[edit]

The reference for in the References section is "Thunderbolt™ Technology". Intel. Retrieved April 30, 2014. In the source, it specifies the publisher as Intel. Should'nt the publisher be set to or something other than Intel? Intel didn't published site, or did it? (talk) 17:57, 7 September 2014 (UTC)

double-digit TB?[edit]

What is "double-digit TB"? -- (talk) 08:13, 10 December 2014 (UTC)

Like the hyperlink under the TB makes clear "Terabyte". And double-digit means what it always does: more than nine – 0-9 being single-digit (lol, rocket science, this ain't)! Given single-digit TB's of data (at this time: 2011, and arguable still is early 2015) is quite a lot for most users, this comment makes clear that anything dealing in double-digits are dealing with more significant volumes. Jimthing (talk) 16:37, 25 March 2015 (UTC)

What are "Apple's external 5K displays"?[edit]

The Thunderbolt 3 section talks about Apple's external 5K displays which don't exist. Apple's only display has 2560 by 1440 pixels[1] and it's 5K iMac can't be used as a display[2]. Duvrai (talk) 21:48, 14 March 2015 (UTC)

Obviously Apple have therefore made 5K displays, just not yet separate ones. TBolt 3 is coming, as are these 4K/5K displays (at 60+Hz, in more affordable cost), this is background on specifications for Tb3 and why it will be needed. Jimthing (talk) 16:27, 25 March 2015 (UTC)
I'm confused as to where you stand. Are you saying that you believe Apple sells external 5k displays? Duvrai just explained that while Apple sells external displays and 5K displays, they don't sell external 5K displays. --Makkachin (talk) 10:28, 22 July 2015 (UTC)


SCSI was developed by Cray (Texas tax payers, Census bureau) to attach multiple boards / cpu in mainframe computing to a central processing unit on the main backplane / bus: or so that the daughter boards could talk independant of the actually central cpu or backplane.

This design (also the new PCI-e), says they can connect multiple devices, full duplex data, multiple hubs, packet talking / intelligent bus controll, to a cpu. The Z80 did that (though the clock speed was low, that was only due to year of production).

Please tell me what any of these "new things" do that the SCSI (say, adaptec AHA-1542) - used in the original apple and available since before ISA bus, does that is any better than SCSI.

I'm beginning to feel it's not just copy-lefting but a rebranding attempt so that foreign countries own what USA had designed: without paying or giving any recognition for. I think "HDMI" and also " becomes Wayland" are good exmaples of that going on. — Preceding unsigned comment added by (talk) 22:59, 16 July 2015 (UTC)


Advertising early on it is an optical design is worthless except as history. There aren't available optical products. Your bolstering a feature which doesn't exist. Any card any day "could use optics". This one doesn't. — Preceding unsigned comment added by (talk) 23:07, 16 July 2015 (UTC)


Is Apple the only manufacturer that sells computers with Thunderbolt support? --Makkachin (talk) 10:23, 22 July 2015 (UTC)

No, i have a gigabyte motherboard with thunderbolt. With the overlap between usb-c and thunderbolt 3, it might even have a chance to have a real future outside of the Apple world. But it may also go down the path of FireWire and the many other standards that only really survive(d) in Apple's ecosystem. PizzaMan (♨♨) 21:54, 11 February 2016 (UTC)

TB3 Invented by "Intel Israel"[edit]

Consensus sought on whether or not it's a useful detail to say it was developed at Intel's Israeli subsidiary. Another editor has twice removed this detail. ... richi (hello) 10:16, 18 September 2016 (UTC)

Interesting fact, sourced: leave in --Zac67 (talk) 11:38, 18 September 2016 (UTC)
  1. What useful information does it lend to the reader? Such mentions are rarely relevant, the exception being major and/or unique facilities like Xerox PARC and Lockheed Martin's Skunk Works. This is hardly the case.
  2. There are hundreds, if not thousands of article about technologies originating in Intel; are we to now search among Intel's dozens of subsidiaries where each of these were developed? And what about Microsoft? Bayer? GE? François Robere (talk) 02:27, 19 September 2016 (UTC)
richi? François Robere (talk) 13:51, 26 September 2016 (UTC)
Imho, Intel Israel holds a special position at least since they developed the Pentium M, the ancestor of all current Intel CPUs, when Intel US was stuck on the ill-designed Netburst line. --Zac67 (talk) 17:27, 26 September 2016 (UTC)
So open Intel Israel and mention it there. All of the above issues still hold. François Robere (talk) 06:07, 27 September 2016 (UTC)