9/17/2014
jcschild
"""As I understand it the lack of PCIe TB cards comes down to Intel's policies and attitudes""""
 
far more than you know.. major politics with this one, they have no issue issuing license to input device manufacturers but to not companies who wish to implement an add on card. and TB 2 onboard implementation is nearly non existent without a headerport and add in card TB 1 was on a lot of mobos.
 
"""a PCI(e) connected card will give much lower latency than USB. Even FW400 often has an edge over USB2 in latency terms,""""
 
not true exactly.  several of RMEs interfaces actually do better on USB than firewire.
Motu FW/USB are identical
RME USB is almost the same as RME PCIe, and RME USB will actually beat a few PCIe cards.
its all about the drivers.
 
 
as to USB 3, RME Motu, Steinberg are fine. others are sketchy, some flat out no go.. (focusrite)
yet to see a desktop not have USB2, some laptops may have only 1 USB2 which could be an issue with multiple USB devices
also depends on which USB3 chipset for the others.. RME, Motu, Steiny never an issue..
It's about the size of the computer. Since I'd like to go forward and use a touch screen I have bought 19" preamps. Wich I now want to connect to a rackmountable pc. It should be as sturdy as possible. Since there now are onboard display chips a thunderbolt connection (like usb and firewire) moves the audio hardware away from the computer. Wich leaves only the driver wich may or may not perform as discribed. 
 
Since Thunderbolt(2) will talk directly to the CPU (in the right MOBO build) it will have nothing to do with the PCI-E. And if a computer only has thunderbolt, HDMI and USB3 it can be small, cool and much less prone to mechanical issues.... And thus survive much better in a rack.... So I thought. 
 
So with two or three 19"rack mounted 8 ch preamps, 1 unit high computer and a touch screen you get this highly portable unit...... Wich appeals....
And to take this discussion to another level..... Behringer bought Klark Teknik and Midas. So they aquired the Sony Oxford built (in 2007 Klark Teknik bought Sony Oxford from Sony) AES50 and AES60. Right now the rest of the industry will move on towards AES67...... But once Mr Behringer finds out it's nothing more than a fancy version of AES60 I'm quite certain he will go to war...... Cause in his mind he bought the future of Pro audio. And left the rest of the bunch to deal with his patents..... 
 
Then there is this thing of matrix within a ethernet connection. And the stuff of what happens when you sum 24 bit audio...... it creates artifacts etc..... (or it may do that....).... When you stay within the DAW system, daw builders can do what they do best.... And process critical audio at much higher specs like 40 bit floating or not or higher....
So Thunderbolt seems to be the best technology (besides PCI-E (but that means putting risercards in motherboards and thus making stuff big measured by nowadays standards). And Thunderbolt was made with break out boxes in mind making additional firewire and usb ports possible. And thus.... Meaning current drivers will (I think.... so this is open for debate here) work with Thunderbolt equipped motherboards....
 
Thunderbolt was offered exclusively to Mac for a couple of years. This has passed. And some issues with the technology arose meaning external graphic cards were put in the freezer. Now Thunderbolt 2 is there. And I think the bottle neck is about a chip between PCI-E and the CPU. Another chip needed on a motherboard wich means money towards another chip and licences to be paid towards INTEL will up the price considderably. So Intel took the time to showcase their technology towards ASUS, MSI, ASROCK and AMD. And half of those manufacturers are chinese. And these guys have not respected patents at all in the past... And because of Thunderbolt2 has arrived the old TB1 did not perform as expected. And because Thunderbolt(2) will not perform as expected as a risercard in a PCI-1 slot, the risercard is simply not a good idea....
 
Well, some stuff I mentioned might be open for debate.... So shoot  \../ 
 
9/18/2014
wst3
Muziekschuur at home
And to take this discussion to another level..... Behringer bought Klark Teknik and Midas. So they aquired the Sony Oxford built (in 2007 Klark Teknik bought Sony Oxford from Sony) AES50 and AES60. Right now the rest of the industry will move on towards AES67...... But once Mr Behringer finds out it's nothing more than a fancy version of AES60 I'm quite certain he will go to war...... Cause in his mind he bought the future of Pro audio. And left the rest of the bunch to deal with his patents.....



A couple minor quibbles...
 
AES-50 is a standard for audio transport over Ethernet, and it predates the IEEE standards that make up AVB. AES50 was extended to become SuperMAC, the Sony developed transport. It was further extended to become HyperMAC, also by Sony. Behringer owns the patents for SuperMAC and HyperMAC, but has no control over the underlying standards.
 
AES-67 is a recent standard, sometimes billed as a standard for audio transport over IP. But is a actually a standard for interconnection between different IP based audio protocols. It references the same IEEE standards that AVB references, but it operates over a higher layer in the network stack.

That said, I think Ethernet and/or IP stand to be the surviving transport for Audio, MIDI, DMX, Video, and Control (there is an AES standards committee hammering out an open control protocol, and there is also an independent consortium working on the same problem. This time around it appears they are talking to each other!)
 
I have a Dante I/O box in my studio. It just works, but latency is not on a par with any of the other approaches... yet. They recently upped the channel count from 64x64 to 512x512, and they are almost certainly working on the latency problem as I type, although I have no insider information one way or the other.
 
The real surprise right now is that Dante (proprietary) is crushing AVB,and Super/HyperMAC in the marketplace. Part of the reason may be audio folks are more comfortable working at the IP layer, Ethernet can be scary (not really) and device discovery/configuration is easier at the IP layer.

A friend of mine and I have played around with generic, network based I/O boxes for audio and MIDI. It will happen, it may not be us. And actually, as fewer and fewer folks depend on MIDI we could be completely off the mark!
One of the reasons I asked about Thunderbolt here is that my attempts to read about it elsewhere seems to have ended in a quagmire of confusion.
 
I have a few questions and hope that maybe people can take a shot and one or more of them.
 
1) With Firewire and Windows there is a basic 1394ohci.sys driver that I have always thought of as a system level driver. I have assumed that this system driver is used so that a firewire appliance can be recognized and this is how the OS is prompted to load the specific device driver that the appliance manufacturer provides with the hardware they sell. Is there a similar paired relationship with Thunderbolt? Has MS provided a system level driver or is the provision of such a driver entirely dependent on third party vendors?
 
2) Am I to understand that Thunderbolt on a Windows CPU system has only been implemented on a motherboard? I haven't searched for a PCI-E thunderbolt add on card... if I do will I find that they do not exist? If that is the case, is it because Intel is preventing third parties from selling such cards?
 
3) Are there two versions of Thunderbolt? Has the first version been replaced by a newer version on MAC? Do Windows users have practical access to both versions or is it the case where Windows users are just getting started with the first version after it has played out in the Mac world?
 
I imagine that some/many/most of these questions are based on gross misunderstandings and as such may seem to reflect said misunderstanding. I'll appreciate any clarification that anyone can offer.
 
Thank You.
 
 
edited grammar
9/19/2014
wst3
Good questions Mike - I'll throw in what little I know... or what I think I know, the literature can be confusing as there are conflicting reports.
 
mike_mccue
1) With Firewire and Windows there is a basic 1394ohci.sys driver that I have always thought of as a system level driver.<snippity>

You pretty much have it right. Windows uses an abstraction layer to separate hardware from software, but of course there has to be an interface. The simple explanation is that you three layers:
1) hardware - the device either sends or receives data in some format under some rules
2) the interface - generally provided by MS - this provides a uniform interface for applications, so for example writing a byte to a port is the same whether it is RS-232, USB, FW, or whatever. Where this is most beneficial is when a port is generic - for example all mice do the same basic thing, report position and motion, so a single interface will work with all mice (class compliant), but if you have dozens of buttons or other tricks you need to write your own driver.
3) the manufacturer provided driver - MS prefers to have these live at the application level, but some specialized hardware requires the driver to live at the interface level.
 
It appears, with respect to TB, that MS has not yet added their low level interface to the OS, and UA is hesitant to write a driver that talks to the hardware.
 
mike_mccue2) Am I to understand that Thunderbolt on a Windows CPU system has only been implemented on a motherboard? I haven't searched for a PCI-E thunderbolt add on card... if I do will I find that they do not exist? If that is the case, is it because Intel is preventing third parties from selling such cards?

Thunderbolt on a PCIe card would be sorta redundant - TB is PCIe, in the same way that eSATA is SATA. So you could have a PCIe card with a TB socket on the rear panel, but the card would not do a whole lot, other than signal conditioning and maybe line protection. The 4 lanes on a TB port are the same as 4 lanes on a motherboard.
 
mike_mccue3) Are there two versions of Thunderbolt? Has the first version been replaced by a newer version on MAC? Do Windows users have practical access to both versions or is it the case where Windows users are just getting started with the first version after it played out in the Mac world?


There are two versions, the initial version provided 10Gb/s of bandwidth, the current version doubles that. I'm pretty sure that is the only difference.
 
Hope this helps...
9/19/2014
Splat
Nice post... Interesting reading...
Thank You Bill,
 The explanations really help put many of the comments regarding Thunderbolt that I encounter in to a context that I can make sense of.
 
 
 
 I have been thinking about Thunderbolt add in cards in terms of using one to add a Thunderbolt interface to an existing computer.
 
 I started thinking like this when MOTU canceled their PCI-E gear a few months ago and replaced it with Thunderbolt gear that has a lower noise floor than it's previous offerings and some really attractive pricing. It seems to me as if people that can make use of Thunderbolt will be seeing some very nice hi analog track count Audio I/O gear introduced at even lower prices than they may be accustomed too, while people depending on PCI-E cards are going to recognize a decreasing selection that will seem increasingly expensive.
9/19/2014
Living Room Rocker
jcschild
1) it most certainly is NOT MS fault please stop with that myth.
it has to do with manufacturers not wanting to write drivers.
case in point Lynx has TB for windows and not only is it real TB (unlike Motu 828x and previously working Apollo that used Firewire protocol thru TB)
and to add to that Lynx can daisy chain (up to 6 units or 192 I/O), all other present TB on the market are end point devices. and look to be for awhile.
 
2) TB is absolutely pointless, it offers NOTHING over what we have now
3) it will NOT give you lower latency
4) it will not offer more bandwidth than PCIe Cards or RME USB3
5) if you own Apple then you don't have much a choice.. (of course this being a sonar forum that would not apply to anyone) but USB still is as good. (assuming the manufacturer knows how to write drivers EG RME)
 
but hey feel free to believe all the marketing hype
 
 


With all due respect, Scott, how is it not on MS but on manufactures when they are already offering full/real TB for Apple machines and not Windows PCs?  And if is not MS, why did Intel go out their way to compensate as described in the article I attached from AnandTech?
 
 "Hanging Thunderbolt peripherals directly off the CPU's PCIe lanes requires extensive support from the operating system, particularly when it comes to hot plugging devices and/or waking up peripherals from sleep mode. Over the PCIe lanes off the PCH, Intel has more control via its chipset drivers. Ultimately, it looks like Microsoft dropped the ball and Intel decided to come up with a certification solution by only allowing Thunderbolt silicon to talk to the PCH for all PC boards.
 
While Microsoft continues to twiddle its thumbs, Intel has decided to come up with less restrictive hardware suggestions to bridge the Thunderbolt experience gap between Macs and PCs."
 
Also, what about the fact the TB can handle multiple protocols whereas USB can only handle USB.  Why overlook the versatility of TB?  Granted it's got some improvement to be desired, but it is the youngest of the bus protocols (if I characterized that correctly).
 
I will take the answers offline.
 
Kind regards,
 
Living Room Rocker
 
P.S.  Will USB3 offer enough bandwidth as TB2 for utilizing a 4K monitor?

 "Hanging Thunderbolt peripherals directly off the CPU's PCIe lanes requires extensive support from the operating system, particularly when it comes to hot plugging devices and/or waking up peripherals from sleep mode. Over the PCIe lanes off the PCH, Intel has more control via its chipset drivers. Ultimately, it looks like Microsoft dropped the ball and Intel decided to come up with a certification solution by only allowing Thunderbolt silicon to talk to the PCH for all PC boards.
 
While Microsoft continues to twiddle its thumbs, Intel has decided to come up with less restrictive hardware suggestions to bridge the Thunderbolt experience gap between Macs and PCs."
 




 
They maybe want to integrate all this into one I/O chip wich will do the active on/off thru switching. And the buffering needed for voltage spiking.
 
Once Windows 9 will be introduced I have a feeling things will change. To implement something so deep into the O.S. is something Microsoft would obviously do in a new O.S.
Wich will be introduced in October.... Right?  Wich probably needs some patching a few times.... So it will take a while. But to have a couple of Mac mini(formfactor) pc cases and to interconnect them with thunderbolt to add CPU power is an interesting thought.
 
No more cd/dvd drives, solidstate drives...., four connector types: thunderbolt (wich can do (brake out box) firewire, usb, HDMI), USB3 ethernet and HDMI.
12
© 2025 APG vNext Commercial Version 5.1

Use My Existing Forum Account

Use My Social Media Account

loading