iMac Pro X299 - Live the Future now with macOS 10.14 Mojave [Successful Build/Extended Guide]

[mm2margaret]

I've been running benches in windows and the card is noisy but it stays very cool compared to Vega 64. The backplate gets a little warm but you can't fry an egg on it like the Vega.

That depends on the Vega.....my Sapphire Nitro Vegas don't get that hot.....but some do.....I used to have one that overheated pretty regularly.

Think I'll wait for the Navi cards (October of this year)....

 

[kgp]

Re: "The BIOS enforced "CSM" by default". Is the same true when booting Windows 10?
If so, something is definitely fouled up with the card's firmware.

Totally unrelated with any OS. With the VII, the BIOS automatically enforces CSM at boot even if one disables CSM in the settings. Which means, if you disable CSM in bios and reboot, system boots but reboots again to enable CSM and clearly states before entering the BIOS splash screen “CSM enabled, GPU not compatible with UEFI”. And for sure, definitely a GPU firmware issue at least in case of the MSI VII.

 

[kgp]

This will be a great card if we can get full support in macOS. Tested in latest beta 14.4 and still no support.
@kgp could you post your ssdt?

Luxmark Ball in Windows is awesome. About double that if Vega 64. Check out this dual setup.
Pentium g3258 $50 with 2X Radeon VIIs.

I edited the 5000, and 5000HW Kexts with device ID but still no acceleration.

View attachment 386102
View attachment 386103

Awesome dude!

Here is the link to my VII SSDT, which apart from a few cosmetically modifications up to now is identical with my Vega64 SSDT but skipping the formerly implemented load table: SSDT-X299-Radeon-VII.aml (for Slot-1 on my ASUS X299 Deluxe). I guess in your case it is not necessary to mention that you might have to adopt ACPI path and ACPI replacements implemented within the SSDT to make it work for your mobo and slot configuration. However, please note that you also need SSDT-DTPG.aml in addition, if method DTPG is not yet defined elsewhere in any of your System SSDTs implemented.

With WE kext + VII SSDT, all 4 ports are properly implemented and fully work including Display Hotplug, multi-monitor support and flawless monitor and system sleep/wake.

Cheers,

KGP

 

[rugula]

Totally unrelated with any OS. With the VII, the BIOS automatically enforces CSM at boot even if one disables CSM in the settings. Which means, if you disable CSM in bios and reboot, system boots but reboots again to enable CSM and clearly states before entering the BIOS splash screen “CSM enabled, GPU not compatible with UEFI”. And for sure, definitely a GPU firmware issue at least in case of the MSI VII.

Try running the free Windows program "GPU-Z", which includes a check box to indicate whether UEFI is supported by the card's firmware, or not.
However, the last GPU-Z version update was Dec. 10, so the Radeon VII card may not yet be supported.
There are other system aspects that could have the same effect of forcing CSM in the bios. If the operating system were installed with CSM enabled & the boot drive using the MBR format, for example, and then afterwards CSM was disabled in the bios. And different motherboards may react to such a configured boot drive in different ways.
My MSI Intel H81 chipset board has two bios options: "Legacy" (CSM enabled) & "Legacy + UEFI" (UEFI is enabled, but can also work with certain legacy options).

 

[kgp]

Well, there are other system aspects that could have the same effect of forcing CSM in the bios. If the operating system were installed with CSM enabled & the boot drive using the MBR format, for example, and then afterwards CSM was switched off in the bios. And different motherboards may react to an identical boot drive in different ways.

Seems you do not understand the problem. The system does not even boot without CSM disabled when the VII is installed, just reboots on boot and automatically enables CSM before its next boot where it finally passes to the BIOS splash screen. There is absolutely nothing one could do about this fact. @DSM2 also confirmed the same BIOS boot issue with the ASUS X299 Sage 10GB and a MSI Radeon VII.

This is not about booting Windows or macOS. This is about booting the system itself. Once CSM is enabled, one is able to set everything under CSM to "UEFI only" or "UEFI first" though and apart from legacy GPU support the rest of the system remains fully UEFI, but CSM is automatically enabled by the BIOS in any case.

No issues with UEFI Clover, or booting Windows or macOS and therefore an issue totally OS unrelated, which affects all users independent from macOS or Windows because it only relates to the initial BIOS system boot. Windows users generally do not mind though to enable CSM and CSM is by default enabled in the BIOS for Windows in any case.

Not necessarily it must be some flaw in the VII firmware. It could be also some actual, general ASUS BIOS firmware incompatibility with the VII firmware, which would require some fixing by ASUS. In any case, let's wait for reports of users with motherboards of different brands.

What for to check with "GPU-Z" under Windows if the VII is UEFI compatible, if it is the X299 Deluxe BIOS which claims VII incompatibly with UEFI and automatically enables CSM for the VII before even passing the Splash screen and booting at all?

 

[Gigamaxx]

Try running the free Windows program "GPU-Z", which includes a check box to indicate whether UEFI is supported by the card's firmware, or not.
There are other system aspects that could have the same effect of forcing CSM in the bios. If the operating system were installed with CSM enabled & the boot drive using the MBR format, for example, and then afterwards CSM was disabled in the bios. And different motherboards may react to such a configured boot drive in different ways.
My MSI Intel H81 chipset board has two bios options: "Legacy" (CSM enabled) & "Legacy + UEFI" (UEFI is enabled, but can also work with certain legacy options).

I was googling CSM on MSI boards and read it was associated with Windows 8.1 enabled/disabled setting? It may be worth testing?

 

[kgp]

I was googling CSM on MSI boards and read it was associated with Windows 8.1 enabled/disabled setting? It may be worth testing?

Well there is not such setting in any ASUS BIOS. There is only an "OS Type" option, which I anyway always change from "Windows" to "Other OS" as the only available alternative.

Do you also face the same CSM BIOS issue and supposed VII UEFI incompatibly claimed by the BIOS on your GIGABYTE Z170M D3H?

With the VII on ASUS motherboards, you do not even reach the BIOS with CSM disabled. The system reboots before even entering the BIOS splash screen, automatically enables CSM, boots and finally also successfully passes the BIOS Splash screen and all other subsequent steps.

 

[rugula]

What for to check with "GPU-Z" under Windows if the VII is UEFI compatible, if it is the X299 Deluxe BIOS which claims VII incompatibly with UEFI and automatically enables CSM for the VII before even passing the Splash screen and booting at all?

The GPU-Z software may act to narrow down the possibilities of where the fault lies: Radeon VII vs. the ASUS motherboard. There are several Youtube reviews of the Radeon VII, and none of those have mentioned being forced to run in CSM enabled mode. Those reviews are not using an Intel X299 chipset board, however.

 

[kgp]

The GPU-Z software may act to narrow down the possibilities of where the fault lies: Radeon VII vs. the ASUS motherboard. There are several Youtube reviews of the Radeon VII, and none of those have mentioned being forced to run in CSM enabled mode. Those reviews are not using an Intel X299 chipset board, however.

Well maybe somebody else can do this test in this case? I performed all yesterdays tests on a system of a friend. I do not own a VII myself yet. But I might get his GPU for further testing purposes the next days. In this case, I could do the test also of course by myself.

 

[AsEvil]

Apparently you did not implement my TB-SSDT for the Deluxe II attached in the other post . The TB ACPI table is not at all properly implemented on your system.

View attachment 386047

Also do you use SSDT-DTPG.aml which is compulsory in addition?

Please use SSDT-X299-TB3HP.aml.zip and SSDT-DTPG.aml.zip once more attached below . This TB-SSDT does not need any additional adaptation for the Deluxe II as I already adopted it for the onboard TTR controller of the latter motherboard.

It should provide the following results:

View attachment 386022

View attachment 386023

Also do you use the following TB BIOS settings?:

TBT Root por Selector                               Thunderbolt USB
Thunderbolt USB Support                             Disabled
Thunderbolt Boot Support                            Disabled
Wake From Thunderbolt(TM Devices)                   Off
Thunderbolt(TM) PCIe Cache-line Size                128
GPIO3 Force Pwr                                     On
Wait time in ms after applying Force Pwr            200
Skip PCI OptionRom                                  Enabled
Security Level                                      SL0-No Security
Reserve mem per phy slot                            32
Reserve P mem per phy slot                          32
Reserve IO per phy slot                             20
Delay before SX Exit                                300
GPIO Filter                                         Enabled
Enable CLK REQ                                      Disabled
Enable ASPM                                         Disabled
Enable LTR                                          Disabled
Extra Bus Reserved                                  65
Reserved Memory                                     386
Memory Alignment                                    26
Reserved PMemory                                    960
PMemory Alignment                                   28
Reserved I/O                                        0
Alpine Ridge XHCI WA                                Disabled

With your Deluxe II, the originally implemented TB ACPI table without any TB-SSDT in your EFI-Folder should look like the one posted by @applemacosxGOD:

View attachment 386036

Yes, I did put SSDT-DTPG.aml and the SSDT-X299-TB3HP.aml you attached into the patched folder in the ACPI folder (as shown in the pictures attached), here are the results of trying your advice.