Positive Thunderbolt thread

[John.Murray]

Are you a Mac or Win user? I still haven’t found any info about TB adoption on the Dark Side (Star Wars humour), and as far as response to my request for info from other members, it’s <sound>[crickets]</sound>.

Is it MS, or timid vendors?

 

Neither - the answer, as usual, has multiple facets:

1) Licensing; Although based on Intel's Light Peak initiative (high speed computer bus implemented on fiber), Thunderbolt, when introduced was owned by Apple.  In July, Apple signed licensing over to Intel, although it's still not clear what rights / licensing restrictions other Vendors utilizing Intel's chipsets will have.

2) Implementation; I'm personally convinced the first release of the forthcoming socket 2011 Z79 Platform will be the Mac Pro.  I expect 1st generation Z79 Mainboard makers to not offer TB.

3) Again, remember that TB is a bus technology - Windows, *nix, support should not be an issue.

[mediumcool]

Hi John

1) Licensing; Although based on Intel's Light Peak initiative (high speed computer bus implemented on fiber), Thunderbolt, when introduced was owned by Apple.  In July, Apple signed licensing over to Intel, although it's still not clear what rights / licensing restrictions other Vendors utilizing Intel's chipsets will have.

As far as I know, Apple registered the trademark; a trademark is not a patent. Intel stressed in May of this year that the IP was freely licensable (they obviously expect to make money on parts); if Intel was bullshitting, who can be trusted? I think Intel and Apple have both learned lessons from the licensing of FireWire. And of course Intel developed USB in the first place.

Found this from February which seems to confirm an exclusivity deal for Apple, which may account for PC-makers not introducing TB until 2012.

2) Implementation; I'm personally convinced the first release of the forthcoming socket 2011 Z79 Platform will be the Mac Pro.  I expect 1st generation Z79 Mainboard makers to not offer TB.

Sorry, but I don’t understand what you are getting at here. Speaking of the Mac Pro, there will be less need for it with the PCI-E expansion possibilities inherent in TB (Magma, Sonnet, AJA so far). I wouldn’t be shocked if it were to be discontinued eventually; Apple sell many more laptops and iMacs (and minis!). Apple seems to be moving in a mobile direction in all parts of its portfolio. Some will not be happy about this.

3) Again, remember that TB is a bus technology - Windows, *nix, support should not be an issue.

Yup, it is a bus technology, but politics, economics and pragmatism collide (and collude ) in engineering decisions; this post raises doubts.

[John.Murray]

I'm sorry - you've misunderstood the intent of my post

1) I have no idea what the exact licensing mechanism is for the "Thunderbolt" moniker, but I do know that Apple formally gave Intel rights on July 9th.  What previous intentions were, are of course moot.  Again the technology had always been Intel's and is implemented on Intel chipsets only.

2) My expectation is that the first non-apple implementations of TB will be on socket 2011 mainboards - I'm sorry if you thought I was implying anything other than that....

3) The ExtremeTech author is guilty of confusing TB with USB 3.0 - they are completely different, TB being a BUS interface.  In point of fact - if you load bootcamp on a TB equipped MacBook Pro, the new Apple TB display will work from Windows 7.....

http://www.anandtech.com/show/4832/the-apple-thunderbolt-display-review/9

The deal is this:  Bus interfaces are implemented at the chipset level and are included in the chipset driver package supplied by the maker.  So, if I buy a newer mainboard and try to install lets say Windows XP on it, I'm going to have to download and install the chipset driver.  In the case of Win 8 - an article talking about what MS will or won't do with a beta O/S thats not even close to Release Candidate stage?  Really?

[mediumcool]

I'm sorry - you've misunderstood the intent of my post

Which one? 

1) I have no idea what the exact licensing mechanism is for the "Thunderbolt" moniker, but I do know that Apple formally gave Intel rights on July 9th.  What previous intentions were, are of course moot.  Again the technology had always been Intel's and is implemented on Intel chipsets only.

This is what I have been trying to find out all along: Intel developed an optical I/O system called Light Peak and was joined in development by Apple, which provided the connector (mini DisplayPort) and very likely some IP to do with transmission over copper rather than optical fibre (Apple has plenty of serial bus expertise, and maybe patents too). So I dispute that the technology has always been Intel’s; it’s been collaborative [at least] in the move to copper. I know that the chipsets are made by Intel, John, as I have referred to in previous posts.

2) My expectation is that the first non-apple implementations of TB will be on socket 2011 mainboards - I'm sorry if you thought I was implying anything other than that....

I have no opinion on what you were implying, or not implying—I simply said that I did not understand it. I am most reluctant to make an inference when I am baffled by a statement. Since I am a Mac user, I have little knowledge or interest in the naming of logic boards (we take what Apple gives us and enjoy it!  ).

3) The ExtremeTech author is guilty of confusing TB with USB 3.0 - they are completely different, TB being a BUS interface.  

I do not see that in the article at all, and USB is (of course) the acronym for Universal Serial Bus; if it looks like a bus, walks like a bus, quacks like a bus etc. it’s a bus! I really don’t know what you’re on about here.

The deal is this:  Bus interfaces are implemented at the chipset level and are included in the chipset driver package supplied by the maker.

Not necessarily so: I have installed USB and FireWire PCI cards on computers which had no hardware support for these serial busses on their main boards. Is it possible that PCI add-in cards with TB connectors will eventually be made available for older computers?

In the case of Win 8 - an article talking about what MS will or won't do with a beta O/S thats not even close to Release Candidate stage?  Really?

Why not? MS should be making clear their intentions vis-a-vis Thunderbolt (only thing I could find yesterday was an old-ish reference to MS testing TB as a networking protocol). In the absence of confirmation, rumours abound.

[mediumcool]

Is it possible that PCI add-in cards with TB connectors will eventually be made available for older computers?

Not alone in this regard.

[mediumcool]

Target Disk booting via Thunderbolt.

EFI is the bottleneck, unfortunately; no real advantage over FW800.

[mediumcool]

Grain of salt required?

[John.Murray]

I do not see that in the article at all, and USB is (of course) the acronym for Universal Serial Bus; if it looks like a bus, walks like a bus, quacks like a bus etc. it’s a bus! I really don’t know what you’re on about here.

The only thing USB, Firewire and PCI-E / TB share is the term BUS - they are not equivalent - sorry.

Not necessarily so: I have installed USB and FireWire PCI cards on computers which had no hardware support for these serial busses on their main boards. Is it possible that PCI add-in cards with TB connectors will eventually be made available for older computers?

Nope, TB add's additional PCI-E lanes, thats one reason why I think first non-apple implementation will be socket 2011.

Why not? MS should be making clear their intentions vis-a-vis Thunderbolt (only thing I could find yesterday was an old-ish reference to MS testing TB as a networking protocol). In the absence of confirmation, rumours abound.

Again, you insist on confusing a true I/O Bus with a peripheral bus - there *is* a huge difference.  Please *look* at the block diagram's in Anand's review of the Apple TB display.  Anand points out that although the Mac TB display works fine running Windows 7, the Promise RAID does not; why?  Because Promise hasn't released a Windows driver for the *device* (yet).  It's no different than plugging an internal RAID card into your computer's PCI-E bus; can your O/S see the card?  Yup.  Can it communicate with the card? - not without a driver.

In short, TB is the *same* as the internal PCI-E Bus on any modern Mainboard -you don't need a driver to be able to communicate through the bus.  So, whenever you read about "product X over Thunderbolt" - substitute in your mind "product X over your PC's internal PCI-E bus"....  

Firewire and USB are designed as external Bus protocols requiring separate chipsets along with a driver to communicate through them.  Both Apple and MS offer native support for these external Bus's (well, Apple not so much for USB 3).  Depending on the device you may very well still need the device driver (scanner, camera, display calibrator).

[mediumcool]

The only thing USB, Firewire and PCI-E and TB share is the term BUS - they are not equivalent - sorry.

Don’t be sorry. They share the name bus because they are busses. Pure/simple; you introduced the word equivalent. And why is USB3 widely seen as a competitor to Thunderbolt, if they are not in some way equivalent in what they offer the user?

Again, you insist on confusing a true I/O Bus with a peripheral bus - there *is* a huge difference.

What is a “false” I/O bus? If it’s outside the computer housing, I am entirely at ease calling it a peripheral, and all TB is after all designed for devices outside the box (prove me wrong, please!). And FireWire under OHCI has had DMA through hardware for a long time. I/O is not the exclusive preserve of one technology.

In short, TB is the *same* as the internal PCI-E Bus on any modern Mainboard -you don't need a driver to be able to communicate through the bus.

John, but what about the logic on the Intel chipset and in the cable? I understand that there are TB drivers for OSX (.kext extensions I assume) that support FB. Maybe somebody with an FB Mac could have a look and post a screenshot.

Your tone seems somewhat aggrieved—I don’t feel that’s necessary. And it should be noted again that I am coming from a Mac perspective, and you from the PC arena. My chief motivation for wanting to understand the slow adoption on the PC side has to do with the viability and availability of products that I would be able to use (I have used Windows in teaching but never for my professional work).

Brings to mind the ironic situation that the iMac [1998] really got USB going, whereas uptake among PC assemblers was modest to begin with.

 

[Farmer]

Why not? MS should be making clear their intentions vis-a-vis Thunderbolt (only thing I could find yesterday was an old-ish reference to MS testing TB as a networking protocol). In the absence of confirmation, rumours abound.

Lol - like Apple always give us so much advance notice about what they intend to do?

When they're ready to share, they will.  It's always been that way and always will be :-)

[mediumcool]

When they're ready to share, they will.  It's always been that way and always will be :-)

Microsoft, or Apple?

[Farmer]

Anyone.

[mediumcool]

Anyone.

OK, but remember the Osborne Effect! 

[John.Murray]

Medium:  Let's look at a working diagram of an X58 chipset board to describe the differences between BUS technologies:  (this image is from wikipedia, open commons license)

Classic Intel architecture has always routed various forms of I/O through 2 primary means:

The "northside" -memory access
The "southside" - all other I/O

Starting with the X58, Intel has moved the "north bus" memory controller on the CPU die itself.  

On the south side, notice where the PCI-E2 Bus lanes originate - this is where Thunderbolt interface is located.  

Compare that with all other I/O through through the ICH10 controller.

What is a “false” I/O bus?  If it’s outside the computer housing, I am entirely at ease calling it a peripheral, and all TB is after all designed for devices outside the box (prove me wrong, please!). And FireWire under OHCI has had DMA through hardware for a long time. I/O is not the exclusive preserve of one technology.

False? I'm differentiating between an Internal and External bus and you know it.  OHCI (an external BUS protocol) does support DMA, but look at the path - it goes through the CPU.  Please read:
http://en.wikipedia.org/wiki/Host_controller_interface
Note the language in the first paragraph regarding the need for a software driver in all implementations?

John, but what about the logic on the Intel chipset and in the cable? I understand that there are TB drivers for OSX (.kext extensions I assume) that support FB. Maybe somebody with an FB Mac could have a look and post a screenshot.

Because of the bandwidth involved, TB cables are active - there is no driver involved with the cable.  Remember this was originally concieved and designed by Intel to be an optical interface.

Your tone seems somewhat aggrieved—I don’t feel that’s necessary. And it should be noted again that I am coming from a Mac perspective, and you from the PC arena. My chief motivation for wanting to understand the slow adoption on the PC side has to do with the viability and availability of products that I would be able to use (I have used Windows in teaching but never for my professional work

I come both both sides, owning Macs and PC's - the architecture between the two is now identical.  I'm not aggrieved at all, but it quite apparent you haven't taken the time to read any of the links I've supplied - Anand's review of Apple's TB display contains a beautiful explanation of what TB is.  Try reading it

[PierreVandevenne]

Brings to mind the ironic situation that the iMac [1998] really got USB going, whereas uptake among PC assemblers was modest to begin with.
 

Yeah, sure.

FWIW, in the end of 1998, (the iMac was announced in May) it was almost impossible to purchase a motherboard without USB... Typical of the time, the 440BX chipset

http://download.intel.com/support/motherboards/desktop/se440bx2/72163201.pdf

http://www.tomshardware.co.uk/440bx-motherboard-review,review-90.html

The reason the start was slow was that USB 1 wasn't really "average user friendly" and did not deliver on the promises (incompatible hubs for example).  USB 1.1 fixed that, in August 1998 and that's this update release that sparked adoption.

The iMac claim to fame is that it dropped other connectors for the keyboard and the mouse. Still, as of today (I still have an early iMac and several 440BX based motherboards) the Mac doesn't recognize early card readers, while the 440BX ones are well behaved (that's a driver issue, but still...).

[mediumcool]

Optical is on its way.

[John.Murray]

Intel finally released presentation notes and content from IDF 2011   Here's a paper talking about TB - hopefully should clear some confusion (in particular page 12)

https://intel.wingateweb.com/us11/scheduler/downloadFileCounting.do?sesfid=BB5EA6E25A8ECAD2E227068168A25C84&abb=AE09539F4DB9D1A900C561F42471AC57&fn=CBCC2DAC2569591FF96D2E557F44CD87D68374B6609B569742BB770ABEE01C5C

[John.Murray]

Acer and Asus announcing Windows based TB devices in 2012

http://www.tomshardware.com/news/Thunderbolt-Light-Peak-Ivy-Bridge-Mooly-Eden-displayPort,13444.html

[julienlanoo]

Hi,

i 've been following this thread, and google-ing around;..
But didn't find any awnsers, maybe you guys would have a better inside..

Is there already on the market a "thunderbolt Hub", for example to USB or to Firewire.
I know sonnet came out with a solution,
and i find loads of posts about a Belkin hub,.. But didn't find anything actually for sale...

So except the Sonnet adapter, is there at this moment already anything for sale out there ?...


thanks
ju

[schrodingerscat]

I purchased a 350MGz G3 Yosemite when they first came out. No floppy, serial, or parallel ports, and had these things called USB and firewire ports, but not much to plug into them at the time. The ADB went the way of the Dodo in the G4.

Everyone yammered on about it the same as now with TB. New technology is built on the back of the previous, remember that old BBC show Connections?

Or as they say - "It steam engines when it's steam engine time".