The only USB 2 device that I got working was my monitor hub, which is powered. However, it is also connected via a USB-C to USB-C cable. I will need to get a simple USB-C to USB-A adapter to do any further testing... (Actually, I jerry-rigged a cable, but got no power on the USB-A side. I'm pretty sure this is because of the lack of pull-downs for power negotiation, though.)
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
MP 1,1-5,1 testing TB3 AIC with MP 5,1
- Thread starter Flint Ironstag
- Start date
-
- Tags
- cmp egpu gigabyte mojave 10.14 tb3
- Sort by reaction score
The only USB 2 device that I got working was my monitor hub, which is powered. However, it is also connected via a USB-C to USB-C cable. I will need to get a simple USB-C to USB-A adapter to do any further testing... (Actually, I jerry-rigged a cable, but got no power on the USB-A side. I'm pretty sure this is because of the lack of pull-downs for power negotiation, though.)
joevt, I saw the thread on https://github.com/intel/thunderbolt-software-user-space/issues/66 that you had posted to (as well as others.)
I brought home the Apple 27" thunderbolt display, and yep... nothing. The card doesn't negotiate anything. No display, ethernet, firewire... nothing.
Am I getting this right that the Alpine Ridge cards will go into legacy mode and support the Apple 27" Thunderbolt display on PC's, but it won't work on classic Mac Pros because the Alpine Ridge cards rely more on the motherboards and BIOS than the Titan Ridge cards. Or will it work, but ONLY give display output?
And the Titan Ridge cards can't talk to the display because the Titan Ridge card can't drop into legacy mode required by the TB1 Apple display. It looks like Gigabyte even resolved the problem and got it to work, but Intel refuses to sign the update, so they can't send it out to fix the card?
Did that thread imply that maybe some variant of Linux may have patched around it, or did I misunderstand that bit?
There was also mention of using an Alpine Ridge dock as an intermediary to get the two to talk... has anyone done that?
I brought home the Apple 27" thunderbolt display, and yep... nothing. The card doesn't negotiate anything. No display, ethernet, firewire... nothing.
Am I getting this right that the Alpine Ridge cards will go into legacy mode and support the Apple 27" Thunderbolt display on PC's, but it won't work on classic Mac Pros because the Alpine Ridge cards rely more on the motherboards and BIOS than the Titan Ridge cards. Or will it work, but ONLY give display output?
And the Titan Ridge cards can't talk to the display because the Titan Ridge card can't drop into legacy mode required by the TB1 Apple display. It looks like Gigabyte even resolved the problem and got it to work, but Intel refuses to sign the update, so they can't send it out to fix the card?
Did that thread imply that maybe some variant of Linux may have patched around it, or did I misunderstand that bit?
There was also mention of using an Alpine Ridge dock as an intermediary to get the two to talk... has anyone done that?
Last edited:
I tried an Apple Keyboard with my GC-TITAN RIDGE. No problem there. It connects to the USB 2.0 hub that I connected the GC-TITAN RIDGE's USB 2.0 header to. There was also a couple devices attached to the keyboard that also worked.
I don't know what the problem is. Maybe DisplayPort output over Thunderbolt 1 is different somehow? I don't know if anyone tried connecting the display to an intermediate device. An intermediate device will cost something like $200 if you want to experiment.
I had the brainstorm of seeing if a 2015 MacBook in target disk mode would act like a Thunderbolt 2 HUB and let me connect the Titan Ridge -> TB3-TB2 -> MacBook -> Thunderbolt display. Nothing happened with the display, nor did the PC see the laptop drive.
However, just for kicks, I switched it up and rebooted the MacBook into MacOS and connected Titan Ridge -> TB3-TB2 -> MacBook (via TB2 to TB2 Cable) and Thunderbolt networking showed active. I assigned a manual IP address to the PC and the Mac and I could ping the PC over thunderbolt. The Mac showed up in the "Attached Devices" Thunderbolt devices panel on the PC.
Now here is the odd bit... I connected Titan Ridge -> TB3-TB2 -> Thunderbolt display (Via TB2 to TB2 cable) -> MacBook (Via the display's cable) and Thunderbolt networking was still working and I could ping the PC. The display lit up as an external display to the MacBook. The Mac was still present in the "Attached Devices" Thunderbolt devices panel on the PC.
Connecting Titan Ridge -> TB3-TB2 -> Thunderbolt display (Via the display's cable) -> MacBook (Via Separate TB2 to TB2 cable) yielded no Thunderbolt networking. But, the display still lit up as an extended desktop of the MacBook.
Is Thunderbolt networking some other protocol? (Yes, found it... it uses the IPoTB protocol.) It did prove that SOME things are making it past the TB3 to TB2 adapter and through the Apple Thunderbolt display.
Maybe Monday I'll bring home a 2018 Macbook and try similar stuff with it.
Other notes: (All of these devices are TB1 - 10Gbit)
Original Apple Thunderbolt Ethernet adapter - Works (Presents as Broadcom NetXtreme Gigabit Ethernet)
Original Apple Thunderbolt Firewire adapter - Works (Presents as LSI 1394 OHCI Host Controller)
G-Disk Thunderbolt drive - Works (Presents as Standard SATA AHCI Controller)
Last edited:
Thunderbolt handles DisplayPort automatically. PCIe is also mostly automatic after some initialization. Other protocols can be added with drivers at both ends of the connection. IP over Thunderbolt is one example that has drivers for Windows, MacOS, and Linux. Apple has protocols for target disk mode and target display mode which don't exist in other OSs. Target disk mode has a driver in EFI (you can't see the drives of a Mac that has booted into MacOS). I believe target display mode has drivers only in MacOS (the Mac has to be booted into MacOS to receive a Target Display Mode signal from another Mac over Thunderbolt).
One question I have is, why can't Target Display Mode (the Thunderbolt version, not the DisplayPort version) work with other Macs that have Thunderbolt (not just the iMac Mid 2011-Mid 2014 models). It seems to me that any target Mac should be able to take the Thunderbolt Target Display Mode signal and dump it to a display unless there's some special hardware that's doing it on the target end. It seems to me that Target Display Mode (Thunderbolt version) would be something like DisplayLink (Display over USB) which takes some resources from the source end to convert a display frame buffer to Thunderbolt. That is a lot more involved than the DisplayPort version of Target Display Mode that is used with the 2009 and 2010 27-inch iMac and therefore might result in dropped frames (or tearing if there's no vertical sync).
https://support.apple.com/en-ca/HT204592
If iMac Thunderbolt Target Display Mode doesn't require extra hardware, then it may be possible to modify the Target Display Mode driver to work with other Macs. I'm not sure that's likely, otherwise why would Apple stop including Thunderbolt Target Display Mode after 2014?
Getting IP over Thunderbolt, Target Disk Mode, and Target Display Mode to work with Macs that have GC-TITAN RIDGE is something to look into after the hot plug issue is solved. Or maybe solving the hot plug issue would solve the other protocols issue.
One question I have is, why can't Target Display Mode (the Thunderbolt version, not the DisplayPort version) work with other Macs that have Thunderbolt (not just the iMac Mid 2011-Mid 2014 models). It seems to me that any target Mac should be able to take the Thunderbolt Target Display Mode signal and dump it to a display unless there's some special hardware that's doing it on the target end. It seems to me that Target Display Mode (Thunderbolt version) would be something like DisplayLink (Display over USB) which takes some resources from the source end to convert a display frame buffer to Thunderbolt. That is a lot more involved than the DisplayPort version of Target Display Mode that is used with the 2009 and 2010 27-inch iMac and therefore might result in dropped frames (or tearing if there's no vertical sync).
https://support.apple.com/en-ca/HT204592
If iMac Thunderbolt Target Display Mode doesn't require extra hardware, then it may be possible to modify the Target Display Mode driver to work with other Macs. I'm not sure that's likely, otherwise why would Apple stop including Thunderbolt Target Display Mode after 2014?
Getting IP over Thunderbolt, Target Disk Mode, and Target Display Mode to work with Macs that have GC-TITAN RIDGE is something to look into after the hot plug issue is solved. Or maybe solving the hot plug issue would solve the other protocols issue.
@Woof Woof, curious. The old ATD should behave similar to the LG display I use. I outlined it earlier in this thread. Try this:
Connect the ATD to an easy to access power-strip or surge protector with an off button. Verify connection of ATD to Thunderbolt 3/2 adaptor to the thunderbolt port on Titan Ridge and verify DisplayPort connections to video card (what video card are you using?). Boot the computer to MacOS or whatever (wait until you know it's started and at login or desktop screen). Turn off the surge protector connected to the ATD. Wait 10-30 seconds. Turn on the surge protector and wait 10-30 seconds. Repeat if necessary. An image should eventually display even if you haven't enumerated via Windows. At this point, it's just using the DisplayPort pass through after the card detected the display signal via power on. To get the thunderbolt, ethernet and USB ports to work on the back, you'll likely need to boot over to Windows...
It *Should* work, but several have tried with no success yet. There was a long thread on github where people were interacting with the devs about the issue and how it was fixed in the lab, but Intel won't sign the update. I think it was joevt that speculated that the Titan Ridge didn't want to communicate with the TB1 controller in the Apple display. Could be that the Titan Ridge can talk directly to TB1 devices, but can't enumerate devices on the other side of a TB1 controller (Pardon any terminology mistakes.) I believe I had read somewhere else that if that handshake to the controller couldn't be made, then DisplayPort wouldn't be output, killing the ability to just use the display ignoring ethernet, firewire, etc.
The github thread mentioned a ton of changes being made to the Linux kernel 5.2 that might impact this. Who knows... maybe something like Ubuntu 19.10 or 20 might light up the display! I don't know enough about Linux to know if the patches they were talking about would be in the 19.10 daily builds, or if they won't show up until 20.
Last edited:
Well, attaching an Alpine Ridge based dock doesn't appear to help with the ATD issue. I've got a Startech TB3DK2DPPD dock, and all interfaces in the dock showed up in both Windows and MacOS (Intel WGI210AT Ethernet, FL1100EX USB 3.0 hub, PCM2912A USB Audio IO) and the displayport connector on the dock worked fine too.
But, the Apple Thunderbolt Display still remained silent.
Still, it is a cool dock with 85w charging. I'll use it at work.
But, the Apple Thunderbolt Display still remained silent.
Still, it is a cool dock with 85w charging. I'll use it at work.
So I just ruled this out. I connected the USB header of the Titan Ridge card directly to the back ports of the machine. Even without the hub in the way, there is no USB 2.0 from the adapters. Strange. Something else is preventing this functionality...
Hi, I've read through this entire thread and I'm still struggling with a few things USB 2.0 related with the GC TITAN RIDGE. I think what you did is brilliant but what is the advantage of using the 4 pin Bluetooth header vs. just adding something like this?
If what I linked could work, do any of the other 9 pin connectors on the hub's board need the be connected to anything and will the GC TITAN RIDGE USB header power it? I really don't understand the signal flow of what's happening here and I'm just outright confused.
USB 2.0 devices connected to the Thunderbolt ports of the GC-TITAN RIDGE are not controlled by the Titan Ridge USB controller. They are controlled by a different controller that you connect to the 9 pin USB 2.0 header of the GC-TITAN RIDGE. You must connect one USB 2.0 port for each Thunderbolt port (two USB 2.0 ports = two 5 pins = one 9-pin). Power comes from the controller end, so don't expect any power from the GC-TITAN RIDGE USB 2.0 header.
Any USB 2.0 source whether it be internal or external should work. There are internal sources for USB 2.0 in the Mac Pro, such as the Bluetooth header or the cable for the front panel connectors. There are external USB 2.0 ports that can be used. Use a hub so that the original purpose of the USB 2.0 source can still work. A 3 port hub is the minimum requirement (one port for the original purpose and two ports for the GC-TITAN RIDGE).
What you linked is two 4 port USB 2.0 hubs that each need to be connected to a USB 2.0 port (two USB 2.0 ports from the 9 pin connector). You can cut it in half for a single 4 port hub (replace the 9 pin with a 5 pin).
Thank you. It was a fun project. I think joevt has expertly addressed your concerns. I just want to emphasize that the point of using the bluetooth header is to provide USB 2 functionality without having to loose any external ports or a PCIe slot.
I was mistaken. 6.1 is the current version not 6.2 and all APFS volumes appear as 'MacOS' in Bootcamp Control Panel as you observed. HPFS volume names are displayed and at the time that is all I had. And your use of brigadier -m iMacPro1,1 is better than using the old message I referenced.
That makes complete sense, thank you. I was mistaken before in thinking that the hub only needed to be connected to the GC TITAN RIDGE USB header (and nothing else). That is why I was confused because that didn't make sense to me how that would just magically allow USB 2.0 to work.
Thank you, I see the point of that now and I think that it's the cleanest method (no wires coming out of cMP case). Also, tapping into the front ports is nice but that's a little more involved. I would probably want to run the 9 pin cable behind the backplane, which is more hassle than it's worth for USB 2.0. I can solder very well and I've been meaning to upgrade my Airport card to BCM943602CD so that will be a good time to add the hub.
Maybe just a little bit. Take the front panel cable, snip the wires for a USB port and insert your hub, similar to what you would do using the Blue Tooth USB. You can get replacement front panel cable from
https://www.powerbookmedic.com/Mac-Pro-Front-Panel-Board-Cable-p-21612-o-3675.html
For maximum flexibility, I would add a 9 pin male connector to the motherboard end for the two USB ports, and two female 5 pins at the front panel end. Then you can decide if you want one or both front ports to come from a hub. If you choose to have both front ports to come from a hub, then you can connect two hubs.
I was able to test USB 2.0 further. I used an Apple USB-C to USB adapter. I tested an Apple keyboard with USB ports. The keyboard seems to work and so does the Apple mouse connected to it:
Code:
USB:
USB 2.0 Bus:
Host Controller Driver: AppleUSBEHCIPCI
PCI Device ID: 0x3a3c
PCI Revision ID: 0x0000
PCI Vendor ID: 0x8086
Hub:
Product ID: 0x2514
Vendor ID: 0x0424 (SMSC)
Version: b.b3
Speed: Up to 480 Mb/sec
Location ID: 0xfa500000 / 1
Current Available (mA): 500
Current Required (mA): 2
Extra Operating Current (mA): 0
Built-In: Yes
Keyboard Hub:
Product ID: 0x1006
Vendor ID: 0x05ac (Apple Inc.)
Version: 96.15
Speed: Up to 480 Mb/sec
Manufacturer: Apple, Inc.
Location ID: 0xfa520000 / 3
Current Available (mA): 500
Current Required (mA): 300
Extra Operating Current (mA): 0
Apple Optical USB Mouse:
Product ID: 0x0304
Vendor ID: 0x05ac (Apple Inc.)
Version: 1.10
Speed: Up to 1.5 Mb/sec
Manufacturer: Mitsumi Electric
Location ID: 0xfa523000 / 7
Current Available (mA): 500
Current Required (mA): 100
Extra Operating Current (mA): 0
Apple Keyboard:
Product ID: 0x0220
Vendor ID: 0x05ac (Apple Inc.)
Version: 0.69
Speed: Up to 1.5 Mb/sec
Manufacturer: Apple, Inc
Location ID: 0xfa522000 / 4
Current Available (mA): 500
Current Required (mA): 20
Extra Operating Current (mA): 0With the Apple USB-C to USB adapter, I was also able to verify that the USB 2.0 data lines are a direct passthrough to the USB header. This is also true for ground but not the 5V pin. Therefore, any powering of devices is entirely left to the CG-TITAN RIDGE.
Last edited:
im at More than 6 months of flawless operation with a Apollo interface.
The only annoyance, manual cold boot.
I have managed to make a apple script to change the startup volume. I can make it also
a "choose startup volume, shutdown" script, instead of manually shutting down.
Now in Windows I found something similar to Automator.
It records the actions of the user.
it works no problem.
BUT,
I didn't find out how to time delay it at start up, so the control panel for bootcamp can load before the script runs.
also I want to be able to omit the script by a shortcut at start up.
other wise its not possible to fix or use Windows, because it shuts down automatically after running the recorded task.
Using only Macs since 93, im lost in windows.
Would be cool if someone could help getting it to work for every one.
How it should work in the end:
Script choosing windows as startup volume and shutting down.
instead of a regular shut down from the menu.
At the next boot, it boot in to Windows, runs the task in windows and reboots
ALL Automatically.
That way the power button has to be pressed ONCE for the whole process,
just like before except that there is a short boot into windows.
I posted that idea several times already.
I don't get it, why keep doing all that manually,
unless there is now a better solution ?
The only annoyance, manual cold boot.
I have managed to make a apple script to change the startup volume. I can make it also
a "choose startup volume, shutdown" script, instead of manually shutting down.
Now in Windows I found something similar to Automator.
It records the actions of the user.
it works no problem.
BUT,
I didn't find out how to time delay it at start up, so the control panel for bootcamp can load before the script runs.
also I want to be able to omit the script by a shortcut at start up.
other wise its not possible to fix or use Windows, because it shuts down automatically after running the recorded task.
Using only Macs since 93, im lost in windows.
Would be cool if someone could help getting it to work for every one.
How it should work in the end:
Script choosing windows as startup volume and shutting down.
instead of a regular shut down from the menu.
At the next boot, it boot in to Windows, runs the task in windows and reboots
ALL Automatically.
That way the power button has to be pressed ONCE for the whole process,
just like before except that there is a short boot into windows.
I posted that idea several times already.
I don't get it, why keep doing all that manually,
unless there is now a better solution ?
Could you share this script, too? Then perhaps we can help you to improve it better
The problem is, is its not a script its a recorded task, most likely it won't work on a other computer.
Any one can record that easily. I didn't even try to write a script in Windows. Im a compleete noob there. other wise I would have shared it along time ago.
If some one could write a script in windows, which triggers that recording, about 10 to 15 seconds after login,
then I could make it work and share all of it as a package with instructions how to set it up.
It needs to be run at boot, but the execution of the screen recording has to be delayed so much,
that it runs after the control panel has loaded the boot champ startup manager icon.
The delay time needs to be customizable in order to work for the whole process as fast as possible.
We need a hot key at start up to omit the execution of the script, incase we have to get in to windows.
Any one can record that easily. I didn't even try to write a script in Windows. Im a compleete noob there. other wise I would have shared it along time ago.
If some one could write a script in windows, which triggers that recording, about 10 to 15 seconds after login,
then I could make it work and share all of it as a package with instructions how to set it up.
It needs to be run at boot, but the execution of the screen recording has to be delayed so much,
that it runs after the control panel has loaded the boot champ startup manager icon.
The delay time needs to be customizable in order to work for the whole process as fast as possible.
We need a hot key at start up to omit the execution of the script, incase we have to get in to windows.
Did anyone try to use Linux as a "helper" to start the TB device? I don't really like the idea of installing Windows, and a small Manjaro system without GUI could be easily used as a chain-like-starter. I know that some folks managed to use the Gigabyte TB on AMD boards, so Linux should be able to initialize the hardware too.
Joevt said it does not work, I never tried, windows on a separate ssd in the second DVD drive slot works fine.
I think the big issue is to automate booting to windows and the reboot to OSX.
Really helpful would be being able to de-activate the automation by key stroke at startup, in case something does not work in windows or if you want to use windows. in my case a keystroke it pretty important because I dont have a boot screen as so many of us.
I think the big issue is to automate booting to windows and the reboot to OSX.
Really helpful would be being able to de-activate the automation by key stroke at startup, in case something does not work in windows or if you want to use windows. in my case a keystroke it pretty important because I dont have a boot screen as so many of us.
Well, my 6 SATA ports are all busy.. ^^
Ordered a mini PCIe-SATA controller for the Airport-slot anyway, not sure if the Mac can boot off an Asmedia controller, if not I swap drives around. I still run a GT 120 and a HD 5770, so boot screen is - yet - no problem. But I'd like to script the (re-)boot process and Linux is way more comfortable for this purpose.
Ordered a mini PCIe-SATA controller for the Airport-slot anyway, not sure if the Mac can boot off an Asmedia controller, if not I swap drives around. I still run a GT 120 and a HD 5770, so boot screen is - yet - no problem. But I'd like to script the (re-)boot process and Linux is way more comfortable for this purpose.
Linux may "sort of" work, but it depends on the situation. I've got an Ubuntu 19.10 install with kernel 5.2, and it can talk to directly attached TB devices like TB to ethernet adapters. However, it won't configure devices attached to Thunderbolt docks/chassis. Oddly, it will SEE the PCIe devices in docks and chassis, but it won't set them up. I tested this with a Startech dock that has an Alpine Ridge controller inside with PCIe Ethernet and PCIe USB3 HUB inside.
So, if you were trying to get some sort of music device to work, it might do, but if you are trying for anything behind another Thunderbolt controller, it likely won't work.
Windows will setup and use all the devices in this same dock after 2 boots, one to enumerate the devices, and a second to set them all up. Of course you can replace the second boot with a boot into HS/Mojave/Catalina and they all see the devices too.