Contribute
Register

GC-ALPINE RIDGE in hands

Status
Not open for further replies.
If the problem is with the iGPU then not using the iGPU would help. I don't know how well mixing iGPU and Nvidia works. I avoid it because iGPU support seems broken in Hackintosh for multiple displays.

The GC-ALPINE RIDGE doesn't have mini-DP output. You need a USB-C DisplayPort alt mode to mini-DP cable for that. The GC-ALPINE RIDGE has the following functions (I don't think any of these functions require the header cable because they all seem to work in macOS on my Z170X-Gaming 7 which doesn't have a Thunderbolt add-in card header):
1) Two 4 lane DisplayPort 1.2 to USB-C DisplayPort alt mode adapters.
2) Two 4 lane DisplayPort 1.2 to Thunderbolt 3 adapters.
3) Dual 4 lane DisplayPort 1.2 to Thunderbolt 3 adapter.
4) Two port USB 3.1 gen 2 controller with 2 lane DisplayPort 1.2 alt mode capability for each port.
5) Thunderbolt 3 NHI (not useable on hackintosh yet?)
6) (I'm ignoring the HDMI output of the GC-ALPINE RIDGE because I don't know what it's good for since it's only HDMI 1.4)

It's the only solution that does all that so I don't think you can call it expensive. The Sunix UPD2018 and Delock 89582 don't have Thunderbolt and only have one DisplayPort input.


Your onboard Thunderbolt controller only gives you one Thunderbolt 3 port and it has only one DisplayPort input so you can't do 5K or two 4K displays, and the input is from the iGPU. The second Thunderbolt 3 port is USB 3.1 gen 2 only with no DisplayPort.

The GC-ALPINE RIDGE has two DisplayPort inputs so you can do 5K or two 4K displays and you can choose to use the IGPU or your graphics card, and it has two Thunderbolt 3 ports. Other than that, there's no performance difference.

The Thunderbolt NVM firmware versions may be different. For example, The GC-ALPINE RIDGE has a newer version than my motherboard. It doesn't look like Gigabyte will ever upgrade my Z170X-Gaming 7 Thunderbolt NVM version. I don't know what version your motherboard has. It doesn't look like there's a firmware update for it yet. Intel should not have given control of the firmware updaters to the motherboard manufacturers because they won't update them.
https://communities.intel.com/thread/111547

@joevt thank you for the detailed reply!

BTW, when I said the Alpine Ridge card was "expensive" I didn't mean it literally—I just meant 'acting as a glorified active video adapter.' In fact, I got it on sale today for only $50 at Fry's Electronics (the card seems hard to find elsewhere in the US)!

Anyways, I just popped in the card so I could send my GTX 1060's video output to my Apple Thunderbolt Display... and it works fine for video, but the other ports on the display (USB/FW/Ethernet, FaceTime camera) are not working in OS X 10.12.6.

Surprising, given that I thought the TB3 via the motherboard header would work the same as the TB3 backpanel port. In Windows 10, when I went to install the Alpine Ridge card's drivers, it was just the same set of Thundebolt drivers I already installed for the backpanel port. Upon launch, the installer only gave me the option to remove the drivers, not to update or install new ones.

If anyone has any ideas, I'd appreciate it!
 
@joevtBTW, when I said the Alpine Ridge card was "expensive" I didn't mean it literally—I just meant 'acting as a glorified active video adapter.' In fact, I got it on sale today for only $50 at Fry's Electronics (the card seems hard to find elsewhere in the US)!
I agree. As an active video adapter, one might imagine that an adapter similar to the Startech or Plugable or Sonnet Thunderbolt 3 to Dual DisplayPort adapter that works in reverse would be a simpler and less expensive implementation since you wouldn't need PCIe, Thunderbolt NHI, or USB support.

Anyways, I just popped in the card so I could send my GTX 1060's video output to my Apple Thunderbolt Display... and it works fine for video, but the other ports on the display (USB/FW/Ethernet, FaceTime camera) are not working in OS X 10.12.6.

The other features of the Thunderbolt Display require PCIe communication over Thunderbolt. Check the Device Manager in Windows (view by connection) and the IORegistryExploer.app in macOS. If you don't see the four Thunderbolt 3 bridges then the card is not being initialized at startup correctly by the bios. On macOS, the PCIe connection has to be made before booting since hot plug is not working for Thunderbolt 3 on Hackintosh yet.

lspci is a simpler alternative to IORegisterExlporer.app since it only lists pci devices. Install lspci V1.1.pkg, reboot (or "sudo kextload lspcidrv.kext" in Terminal.app). Before using lspci use the "update-pciids" command to update the pci id database. Then use "lspci -nnvt" command to dump the PCI device tree.

What Thunderbolt 3 to Thunderbolt 2 adapter are you using? I think the Apple adapter requires a newer NVM (18) version than the Startech or other adapters. I think the GC-ALPINE RIDGE card comes with NVM 21 so that shouldn't be a problem? My motherboard has NVM 16, but the Apple Adapter seems to work anyway

Surprising, given that I thought the TB3 via the motherboard header would work the same as the TB3 backpanel port.
I think you mean TB3 via motherboard PCIe slot, since the header has very little to do with Thunderbolt functionality. The header has the following pins:
1: Force Power signal. I think ASUS has a BIOS option for this which comes from the SuperIO chip. I think Gigabyte uses a pull up resister (default) or a pull down resister.
2: Plug Event signal to a GPIO input of the PCH.
3: S3 signal from the motherboard used by many devices for sleep.
4: S4/S5 signal from the motherboard (usually S5) used by many devices for deeper sleep.
5: Ground
Thunderbolt 3 can mostly work without sleep support, and without plug event.

In Windows 10, when I went to install the Alpine Ridge card's drivers, it was just the same set of Thundebolt drivers I already installed for the backpanel port. Upon launch, the installer only gave me the option to remove the drivers, not to update or install new ones.
If anyone has any ideas, I'd appreciate it!
What about the firmware updater for the GC-ALPINE RIDGE? I suppose the card comes with the same version as the version Gigabyte has on the GC-ALPINE RIDGE support web page (NVM 21)? Post a screenshot from Windows 10 of the Thunderbolt Details window (Thunderbolt Software -> Settings -> Details), and the Attached Thunderbolt Devices window.

You could ignore Gigabyte's firmware updater, and use the Intel NUC Thunderbolt 3 Firmware Update Tool Version 25 (it seems to be the newest?), and point it at the Thunderbolt controller and select the corresponding bin file. The firmware updater checks the bin file to make sure it matches the features of the Thunderbolt controller that you point it to. This means you can't use a newer firmware bin file if it is does not contain the identifiers for your Thunderbolt controller. The file name of the bin file may contain many of the identifiers (manufacturer, chip model, number of ports, etc.). The Firmware Update Tool has command lines to query the Thunderbolt controller's current NVM version which you can use in a Windows cmd window (run as administrator probably).
 
I agree. As an active video adapter, one might imagine that an adapter similar to the Startech or Plugable or Sonnet Thunderbolt 3 to Dual DisplayPort adapter that works in reverse would be a simpler and less expensive implementation since you wouldn't need PCIe, Thunderbolt NHI, or USB support.



The other features of the Thunderbolt Display require PCIe communication over Thunderbolt. Check the Device Manager in Windows (view by connection) and the IORegistryExploer.app in macOS. If you don't see the four Thunderbolt 3 bridges then the card is not being initialized at startup correctly by the bios. On macOS, the PCIe connection has to be made before booting since hot plug is not working for Thunderbolt 3 on Hackintosh yet.

lspci is a simpler alternative to IORegisterExlporer.app since it only lists pci devices. Install lspci V1.1.pkg, reboot (or "sudo kextload lspcidrv.kext" in Terminal.app). Before using lspci use the "update-pciids" command to update the pci id database. Then use "lspci -nnvt" command to dump the PCI device tree.

What Thunderbolt 3 to Thunderbolt 2 adapter are you using? I think the Apple adapter requires a newer NVM (18) version than the Startech or other adapters. I think the GC-ALPINE RIDGE card comes with NVM 21 so that shouldn't be a problem? My motherboard has NVM 16, but the Apple Adapter seems to work anyway


I think you mean TB3 via motherboard PCIe slot, since the header has very little to do with Thunderbolt functionality. The header has the following pins:
1: Force Power signal. I think ASUS has a BIOS option for this which comes from the SuperIO chip. I think Gigabyte uses a pull up resister (default) or a pull down resister.
2: Plug Event signal to a GPIO input of the PCH.
3: S3 signal from the motherboard used by many devices for sleep.
4: S4/S5 signal from the motherboard (usually S5) used by many devices for deeper sleep.
5: Ground
Thunderbolt 3 can mostly work without sleep support, and without plug event.


What about the firmware updater for the GC-ALPINE RIDGE? I suppose the card comes with the same version as the version Gigabyte has on the GC-ALPINE RIDGE support web page (NVM 21)? Post a screenshot from Windows 10 of the Thunderbolt Details window (Thunderbolt Software -> Settings -> Details), and the Attached Thunderbolt Devices window.

You could ignore Gigabyte's firmware updater, and use the Intel NUC Thunderbolt 3 Firmware Update Tool Version 25 (it seems to be the newest?), and point it at the Thunderbolt controller and select the corresponding bin file. The firmware updater checks the bin file to make sure it matches the features of the Thunderbolt controller that you point it to. This means you can't use a newer firmware bin file if it is does not contain the identifiers for your Thunderbolt controller. The file name of the bin file may contain many of the identifiers (manufacturer, chip model, number of ports, etc.). The Firmware Update Tool has command lines to query the Thunderbolt controller's current NVM version which you can use in a Windows cmd window (run as administrator probably).

Thanks again for the details, and the pinout explanation of the mobo header.

Yes, I'm using the Apple TB3-2 adapter. Here are the screenshots as requested from Windows 10's Device Manager and the Thunderbolt Software.

In Device Manager, I don't see four 'lanes' per se; just one Thunderbolt controller. However, all the ports on the display are visible in Device Manager (Firewire/IEEE 1394 controller, Ethernet port, USB hub, etc). They do work in Windows 10.

In the Thunderbolt Software window, I don't have a "details" tab, and it doesn't show any firmware versions; just the connected display.

When booted in OS X, I have the attached in IORegistryExplorer. If I'm understanding it correctly, it does seem to show four lanes, of which one is connected (pci-bridge@0). Is that correct? The only device plugged into the Alpine Ridge card is the Apple TB Display (via TB3-2 adapter).

thunderbolt card in device manager 2.PNG
thunderbolt attached.png
IORegistryExplorer TB3 card.png
 
Yes, I'm using the Apple TB3-2 adapter.
It should work. I wonder if the other adapters will have different behavior (probably not)? They are more expensive than Apple's, are not as compact, and don't work in reverse.
In Device Manager, I don't see four 'lanes' per se; just one Thunderbolt controller. However, all the ports on the display are visible in Device Manager (Firewire/IEEE 1394 controller, Ethernet port, USB hub, etc). They do work in Windows 10.
I didn't say four lanes. I said four PCI bridges. The Device Manager is missing some information. Consider updating the Intel Chipset drivers to the latest. Then you will see Root Port numbers instead of just "PCI-to-PCI Bridge" under the PCI Express Root Complex.
https://downloadcenter.intel.com/do...ice-Software-INF-Update-Utility-?product=1145
When booted in OS X, I have the attached in IORegistryExplorer. If I'm understanding it correctly, it does seem to show four lanes, of which one is connected (pci-bridge@0). Is that correct? The only device plugged into the Alpine Ridge card is the Apple TB Display (via TB3-2 adapter).
Correct. The GC-ALPINE RIDGE card is connected to Root Port 21 according to IORegistryExplorer.app. A bridge consists of two parts, an Upstream Port and a Downstream Port. A Root Port is like an Upstream Port. The Thunderbolt 3 chip has these four PCI bridges:
1) @0 The first bridge has the "Thunderbolt(TM) Controller - 1577". This is the NHI.
2) @1 The next PCI-to-PCI Bridge is port 1 (for PCIe communication). Maybe try connecting the Thunderbolt display to port 1 instead of port 2?
3) @2 The third PCI-to-PCI Bridge is for the USB 3.1 XHC. This is missing from your Windows and MacOS output. Strange. Maybe try without connecting the header cable? Is the Intel USB 3.1 Driver installed (from the GC-ALPINE RIDGE support page). This seems like an issue you should talk to Gigabyte tech support about. Can you connect a USB device to the Thunderbolt 3 port? You should definitely make sure USB works from both ports.
4) @4 The fourth PCI-to-PCI Bridge is port 2 (for PCIe communication). This is where your Thunderbolt display is currently connected. It is missing from macOS. This is why I suggest trying port 1.

Basically, the layout of the GC-ALPINE RIDGE should match the layout of your motherboard's Thunderbolt 3 chip, which you should show in your screen shot for comparison purposes.

The lspci output would be useful since it will also show all the bus numbers (and vendor/device ids). Maybe macOS is running out of bus numbers? 255 is the usual max (for server type motherboards like x99? my old Mac Pro 2008 says 255), but it might be limited to 127 (for Z170, Z270 motherboards?). The MCFG ACPI table has that info which you can see using MaciASL.app. The End Bus Number could be as low as 63, but that's not likely on systems supporting Thunderbolt.
In the Thunderbolt Software window, I don't have a "details" tab, and it doesn't show any firmware versions; just the connected display.

I meant the Details window, when you click the "Details..." button in the Settings window which is opened by right-clicking the "Thunderbolt™ Software" task bar icon and selecting "Settings".

In the "Settings" window, in the Device Connection tab, make sure "Only allow Thunderbolt Devices that have been certified for PC." is not selected.
 
@joevt I installed the Intel chipset drivers as you advised, and it does indeed now show port numbers in Windows.

Strangely, if you look at the two ports listed in the Thunderbolt software details (attached), one listed controller has a much older NVM firmware version than the other. Is the older controller my mobo's built-in port?

The other strange aspect is that they have different security level settings even though I chose "no security" in the BIOS.

I also switched the cables in and out of the card to use port 1. The IORegistryExplorer view in macOS looked the same as earlier, and none of the display's FW/Ethernet/USB ports appeared; I'll install lspci and check that out too.

Strangely, when I enabled "Thunderbolt Boot Support" in the BIOS, macOS booted with some apparent complaints, and very slow/intermittent USB support for my keyboard and mouse, but upon a second restart with that BIOS option disabled, some of the display features seemed to work. FireWire, Ethernet and the USB hub showed up in System Information, but the FaceTime camera and display audio (which I thought were both wired internally using USB) did not work.

Could you enlighten me as to your BIOS's Thunderbolt settings? Am I supposed to be able to independently change the card and onboard TB settings?

Thanks again for all your help.



image.png image.png image.png


It should work. I wonder if the other adapters will have different behavior (probably not)? They are more expensive than Apple's, are not as compact, and don't work in reverse.

I didn't say four lanes. I said four PCI bridges. The Device Manager is missing some information. Consider updating the Intel Chipset drivers to the latest. Then you will see Root Port numbers instead of just "PCI-to-PCI Bridge" under the PCI Express Root Complex.
https://downloadcenter.intel.com/do...ice-Software-INF-Update-Utility-?product=1145

Correct. The GC-ALPINE RIDGE card is connected to Root Port 21 according to IORegistryExplorer.app. A bridge consists of two parts, an Upstream Port and a Downstream Port. A Root Port is like an Upstream Port. The Thunderbolt 3 chip has these four PCI bridges:
1) @0 The first bridge has the "Thunderbolt(TM) Controller - 1577". This is the NHI.
2) @1 The next PCI-to-PCI Bridge is port 1 (for PCIe communication). Maybe try connecting the Thunderbolt display to port 1 instead of port 2?
3) @2 The third PCI-to-PCI Bridge is for the USB 3.1 XHC. This is missing from your Windows and MacOS output. Strange. Maybe try without connecting the header cable? Is the Intel USB 3.1 Driver installed (from the GC-ALPINE RIDGE support page). This seems like an issue you should talk to Gigabyte tech support about. Can you connect a USB device to the Thunderbolt 3 port? You should definitely make sure USB works from both ports.
4) @4 The fourth PCI-to-PCI Bridge is port 2 (for PCIe communication). This is where your Thunderbolt display is currently connected. It is missing from macOS. This is why I suggest trying port 1.

Basically, the layout of the GC-ALPINE RIDGE should match the layout of your motherboard's Thunderbolt 3 chip, which you should show in your screen shot for comparison purposes.

The lspci output would be useful since it will also show all the bus numbers (and vendor/device ids). Maybe macOS is running out of bus numbers? 255 is the usual max (for server type motherboards like x99? my old Mac Pro 2008 says 255), but it might be limited to 127 (for Z170, Z270 motherboards?). The MCFG ACPI table has that info which you can see using MaciASL.app. The End Bus Number could be as low as 63, but that's not likely on systems supporting Thunderbolt.

I meant the Details window, when you click the "Details..." button in the Settings window which is opened by right-clicking the "Thunderbolt™ Software" task bar icon and selecting "Settings".

In the "Settings" window, in the Device Connection tab, make sure "Only allow Thunderbolt Devices that have been certified for PC." is not selected.
 

Attachments

  • image.png
    image.png
    86.6 KB · Views: 597
@joevtStrangely, if you look at the two ports listed in the Thunderbolt software details (attached), one listed controller has a much older NVM firmware version than the other. Is the older controller my mobo's built-in port?
Your motherboard uses the C-Step Thunderbolt 3 controller which uses a different NVM version. The GC-ALPINE RIDGE and my Z170 motherboard uses the B-Step Thunderbolt 3 controller.
Code:
1575    DSL6340 Thunderbolt 3 NHI (B step) [Alpine Ridge 2C 2015]       NVM 21
1577    DSL6540 Thunderbolt 3 NHI (B step) [Alpine Ridge 4C 2015]       NVM 21
15d9    JHL6340 Thunderbolt 3 NHI (C step) [Alpine Ridge 2C 2016]       NVM 9
15d2    JHL6540 Thunderbolt 3 NHI (C step) [Alpine Ridge 4C 2016]       NVM 9
15bf    JHL6240 Thunderbolt 3 NHI (Low Power) [Alpine Ridge LP 2016]    NVM 4

The other strange aspect is that they have different security level settings even though I chose "no security" in the BIOS. Could you enlighten me as to your BIOS's Thunderbolt settings? Am I supposed to be able to independently change the card and onboard TB settings?
Yes, the BIOS is kind of dumb. I'm not sure what to do about that. Tell Gigabyte to fix support for the GC-ALPINE RIDGE card on your motherboard. On my motherboard (GA-Z170X-Gaming 7), the GC-ALPINE RIDGE card causes some kind of conflict in Windows (this motherboard does not have a Thunderbolt addin card header so it's not supposed to be able to support the add-in card).

My Thunderbolt BIOS settings are these:
Code:
Thunderbolt(TM) Support : Enabled
Thunderbolt Boot Support : Enabled
ThunderBolt Detachable Graphics Support : Enabled (maybe I should disable this since I don't use egfx)
    Reservered Bus for Dgpu : 8 (not editable)
    Reserved Memory for Dgpu : 32 (not editable)
    Reserved PMemory for Dgpu : 256 (not editable)
    Select card for Egfx : Select AMD/NVIDA both ogpu
Skip PCI OptionRom : Disabled
Security Level : No Security
Reserved Memory : 256
Reserved PMemory : 32

I also switched the cables in and out of the card to use port 1. The IORegistryExplorer view in macOS looked the same as earlier, and none of the display's FW/Ethernet/USB ports appeared; I'll install lspci and check that out too.
Yes please.

Strangely, when I enabled "Thunderbolt Boot Support" in the BIOS, macOS booted with some apparent complaints, and very slow/intermittent USB support for my keyboard and mouse, but upon a second restart with that BIOS option disabled, some of the display features seemed to work. FireWire, Ethernet and the USB hub showed up in System Information, but the FaceTime camera and display audio (which I thought were both wired internally using USB) did not work.
It would be helpful or at least informative to have a dump of your ioreg from IORegistryExplorer in that case. The Thunderbolt display uses a USB controller for the camera and audio probably. The ioreg dump would have information about how the USB is working. What version of macOS are you using?

In one of your screenshots, it looks like Root Port 5 has the motherboard Thunderbolt 3 chip, but it's missing the NHI, the XHC, and port 2 bridge. I think you need to try getting the motherboard Thunderbolt 3 controller working properly first, before installing the GC-ALPINE RIDGE. Maybe the C-step chip has different behavior than the B-step that we're used to. But you should be able to use a USB device from your motherboard's Thunderbolt 3 chip.
 

I think you need to get it working in Windows before you start trying it in macOS. Try talking to Gigabyte technical support.

The Thunderbolt driver is supposed to be installed first. Then instal the FW Update Tool.

Maybe you need a Thunderbolt 3 device to be connected first, then boot into Windows. Then try the FW Update Tool.

Check the Device Manager. View by Connection. What does it look like? The root port for the PCH PCIEX4 slot should have four bridges: NHI, port 1, USB 3.1 controller, port 2.

Try connecting a USB device into the Thunderbolt 3 port?

Maybe there's a problem with the motherboard or add-in card? Try a different motherboard. Try a different GC-ALPINE RIDGE card. Make sure the header cable is connected correctly to the correct header on the motherboard. Or maybe boot without the header cable connected?


So I got it to work. It came down to user error which is probably true for most things.

I admit that I'm not the most knowledgable about custom computers, hackintoshes, etc. My mistake was having a second M.2 drive installed which was using the PCIEX4 slot. I did not realize this was the case. So I pulled the drive and presto, the card works perfectly. I am not sure why the card worked at all with the m.2 drive in. This led to my confusion.

Next step is figuring out where my windows install shall live but for now I got my external drive working with the display.

Maybe my mistake will help someone else.

Thank you very much for all your help joevt!
 
My mistake was having a second M.2 drive installed which was using the PCIEX4 slot. I did not realize this was the case. So I pulled the drive and presto, the card works perfectly.

Glad you got it working. I should have asked about M.2 when I asked about the PCIEX4 slot. This issue is mentioned in the manual. My Z170X-Gaming 7 motherboard manual says "The PCIEX4 slot shares bandwidth with the M2H_32G connector. The PCIEX4 slot will become unavailable when an SSD is installed in the M2H_32G connector." Your Z270X-Gaming K7 motherboard manual says "The PCIEX4 slot shares bandwidth with the M2P_32G connector. The PCIEX4 slot will operate at up to x2 mode when a PCIe SSD is installed in the M2P_32G connector." So in your case, the Thunderbolt 3 chip could have used x2 instead of x4, but I guess the firmware says not to bother? Maybe it could do it if the firmware were smarter. A Thunderbolt 3 chip can do up to 22
Gbps of PCIe communication (according to https://thunderbolttechnology.net/sites/default/files/Thunderbolt3_TechBrief_FINAL.pdf ). That's about 2750 MB/s (decimal) or 2685 MiB/s. The LaCie Bolt3 says up to 2800 MB/s (I think that's decimal, does anyone use binary MiB for disk speeds?). An x2 connection could only do up to 1970 MB/s. Some motherboards are designed with two PCIe lane Thunderbolt 3 controllers. An x4 connection should allow up to 3940 MB/s (for example some M.2 NVMe SSDs get up 3500 MB/s) so I don't know why more speed isn't allowed?
https://www.dell.com/support/articl...underbolt-3-40gbps-data-transfer-rate?lang=en

I am not sure why the card worked at all with the m.2 drive in. This led to my confusion.
Like I said then, DisplayPort conversion to Thunderbolt 3 is automatic. The add-in card only requires power. It does not require an actual PCIe connection to your motherboard which it can't get when you have the second M.2 slot occupied in your Z270X-Gaming K7 motherboard or my Z170X-Gaming 7 motherboard. I moved my Windows install to a 2.5" SSD. My other option would be a PCIe card with M.2, but then my graphics card would need to run at x8 (which doesn't effect performance noticeably).
 
It would be helpful or at least informative to have a dump of your ioreg from IORegistryExplorer in that case. The Thunderbolt display uses a USB controller for the camera and audio probably. The ioreg dump would have information about how the USB is working. What version of macOS are you using?

In one of your screenshots, it looks like Root Port 5 has the motherboard Thunderbolt 3 chip, but it's missing the NHI, the XHC, and port 2 bridge. I think you need to try getting the motherboard Thunderbolt 3 controller working properly first, before installing the GC-ALPINE RIDGE. Maybe the C-step chip has different behavior than the B-step that we're used to. But you should be able to use a USB device from your motherboard's Thunderbolt 3 chip.

Before I installed the Alpine Ridge card, the motherboard's Thunderbolt 3 was indeed working properly (on both Windows 10 and macOS 10.12.6). I could use the Thunderbolt Display and all its ports/functions from both OSes, with the caveat that its video was only from the iGPU. I was hoping that the stability issues (random shutdown/reboots, and inability to sleep) would be fixed with the standalone TB3 card, since I could power both displays from my GTX 1060, and stop using the iGPU entirely.

Last night I tried a combination of your suggestions and the SSDT hotpatching approach from @dracon . I was able to get the machine booting, but at first the Thunderbolt display features still didn't work.

Yet once I detached the motherboard header cable from the Alpine Ridge card and rebooted, the Thunderbolt Display features (Ethernet, Firewire, USB camera/mic, etc) all started working! I then rebooted into Win10 and all was still working (the Settings/Details in the Thunderbolt Software and Device Manager looked identical as before I detached the header cable). I then put the machine to sleep in Windows overnight.

However, when restarting it this morning into macOS, it was back to the old functionality (where video output over TB worked, but none of the USB-based features).

A second restart into macOS, and all USB features came back.

Very strange. I can't say I understand it. The times when it hasn't worked, I did see some complaints during boot about the Broadcom Ethernet chip used in the monitor (AppleBCM5701Ethernet WaitForFirmwareInit - timeout, GrcReset, getAdapterInfoFailed, etc). But again, on restart these go away.

On wake in macOS, both monitors had a lot of flickering display corruption. Some of that might be due to the GTX 1060 having flaky web drivers (going to display preferences and clicking 'detect' stopped the flickering), but one curious aspect is that all the Thunderbolt Display's ports stopped working again after waking from sleep. Is this a byproduct of the mobo TH3 header containing a 'sleep' pin, as you mentioned before?

I plan to sanity check the USB ports I'm using in the SSDT patch as a next step, but as always, any advice would be great. Thanks!
 
Glad you got it working. I should have asked about M.2 when I asked about the PCIEX4 slot. This issue is mentioned in the manual. My Z170X-Gaming 7 motherboard manual says "The PCIEX4 slot shares bandwidth with the M2H_32G connector. The PCIEX4 slot will become unavailable when an SSD is installed in the M2H_32G connector." Your Z270X-Gaming K7 motherboard manual says "The PCIEX4 slot shares bandwidth with the M2P_32G connector. The PCIEX4 slot will operate at up to x2 mode when a PCIe SSD is installed in the M2P_32G connector." So in your case, the Thunderbolt 3 chip could have used x2 instead of x4, but I guess the firmware says not to bother? Maybe it could do it if the firmware were smarter.

I have a question about your PCIEX4 and M2P_32G slots. What root port does each use when they are both by themselves (x4)? What root port does each use when they are both used (x2, x2)? I wonder if the GC-ALPINE RIDGE doesn't work as x2 because the root port changed? I think I've seen that my GC-ALPINE RIDGE can show up as x1 but I have to reproduce that to be sure.

On wake in macOS, both monitors had a lot of flickering display corruption. Some of that might be due to the GTX 1060 having flaky web drivers (going to display preferences and clicking 'detect' stopped the flickering), but one curious aspect is that all the Thunderbolt Display's ports stopped working again after waking from sleep. Is this a byproduct of the mobo TH3 header containing a 'sleep' pin, as you mentioned before?
I don't know how the Thunderbolt controller uses the information from the TH3 header. Disconnecting the header cable does something - but I can't be sure that leaving the pins floating is a good idea - hopefully internal pull up or pull down resisters are used to select proper default levels. Another option is to add your own pull up (to 3.3V) or pull down (to ground) resister to certain pins of the header of the GC-ALPINE RIDGE - but I don't know what they all do exactly. I think 10KΩ pull up for force power, S3, and S4/S5 doesn't have bad effects but I don't know if it has good affects.
 
Status
Not open for further replies.
Back
Top