MP 1,1-5,1 Experience with MacVidCards Europe RX 6600 XT - no bootscreen

[Dayo]

Now I'm looking for how to rigidly set the resolution through OCLP to 1920x1200:1

I think OC BootPicker has option for setting the resolution to a different mode but it has to be a mode that is provided by the EFI driver.

Maybe you can ask Marek if the GPU has GOP or UGA or both.

RefindPlus might be useful here as you can easily set a target resolution (will only work if available though) and the debug log will tell whether you have GOP and/or UGA. It will always pick GOP if available and fallback on UGA if available when GOP is not. In either case, availability status of both is logged (Detailed for GOP ... Available modes and resolutions etc).

 

[jsts]

I think there's more info you can get. Change DisplayLevel to 2147483714. Attach the logs as zipped archive.

Found EFI/OC/config.plist:

<key>Debug</key>
        <dict>
            <key>AppleDebug</key>
            <true/>
            <key>ApplePanic</key>
            <true/>
            <key>DisableWatchDog</key>
            <false/>
            <key>DisplayDelay</key>
            <integer>0</integer>
            <key>DisplayLevel</key>
            <integer>2147483650</integer>
            <key>LogModules</key>
            <string>*</string>
            <key>SysReport</key>
            <false/>
            <key>Target</key>
            <integer>3</integer>
        </dict>

and then modified to:

<key>DisplayLevel</key> <integer>2147483714</integer>

Reboot. But I did not find any significant changes in the log (attached opencore-2022-08-07-094644.txt.zip).

Try without sudo.
/System/Library/Extensions/AppleGraphicsControl.kext/Contents/MacOS/AGDCDiagnose -a > ~/Downloads/AGDCDiagnose.txt

It works! I attach the AGDCDiagnose output: AGDCDiagnose.txt.zip

 

[jsts]

Right, but this is not "every monitor on the planet" it is one of few Apple monitors. How many monitors Apple sells that they can't be all tested?

I agree with your viewpoint. Considering that the promotional text and technical details does not specify any non-functionality exceptions, the whole thing seems odd to say the least. Hopefully, the MVC-guys can fix this annoying flaw in the future.

 

[startergo]

Maybe you can ask Marek if the GPU has GOP or UGA or both

I think the right person to ask is @netkas, because he is the developer.

 

[jsts]

Or try AllRez.

AllRez output attached (AllRez-Apple-LED-Cinema-Display-24.txt.zip)

 

[jsts]

I'm trying to get the OpenCore bootscreen/bootpicker to be displayed (now the MVC rx6600xt with Apple LED Cinema Display 24" has a black screen). I make changes in EFI/OC/config.plist , "Output" section:

Original OCLP:

<key>Output</key>
        <dict>
            <key>ClearScreenOnModeSwitch</key>
            <false/>
            <key>ConsoleMode</key>
            <string></string>
            <key>DirectGopRendering</key>
            <false/>
            <key>ForceResolution</key>
            <false/>
            <key>GopPassThrough</key>
            <string>Disabled</string>
            <key>IgnoreTextInGraphics</key>
            <false/>
            <key>ProvideConsoleGop</key>
            <true/>
            <key>ReconnectGraphicsOnConnect</key>
            <false/>
            <key>ReconnectOnResChange</key>
            <false/>
            <key>ReplaceTabWithSpace</key>
            <false/>
            <key>Resolution</key>
            <string>Max</string>
            <key>SanitiseClearScreen</key>
            <false/>
            <key>TextRenderer</key>
            <string>BuiltinGraphics</string>
            <key>UIScale</key>
            <integer>-1</integer>
            <key>UgaPassThrough</key>
            <false/>
        </dict>

Changed to:
<key>Resolution</key> <string>1920x1200</string>

Result log:

05:465 00:018 OCC: GOP exists on ConsoleOutHandle and has 1 modes
05:487 00:021 OC: Requested resolution is 1920x1200@0 (max: 0, force: 0) from 1920x1200
05:508 00:021 OCC: Requesting 1920x1200@0 (max: 0) resolution, curr 0, total 1
05:530 00:021 OCC: Current FB at 0x80000000 (0x300000), format 1, res 1024x768 scan 1024
05:551 00:020 OCC: Mode 0 - 1024x768:1
05:573 00:021 OCC: No compatible mode for 1920x1200@0 (max: 0) resolution
05:591 00:018 OC: Changed resolution to 1920x1200@0 (max: 0, force: 0) from 1920x1200 - Not Found
05:611 00:019 OCC: Using builtin text renderer with 1 scale
05:633 00:022 OCC: Install console control (753B79C0/0/0), current - Success
05:654 00:020 OCC: Setup ASCII Output - Success
05:676 00:022 OC: Requested console mode is 0x0 (max: 0) from
05:697 00:020 OC: Requested not to use audio
05:717 00:020 OC: OcMiscLoadSystemReport...
05:736 00:019 OC: OcLoadAcpiSupport...

Changed to:
<key>Resolution</key> <string>1280x800</string>

Result log:

05:617 00:021 OCC: GOP exists on ConsoleOutHandle and has 1 modes
05:639 00:021 OC: Requested resolution is 1280x800@0 (max: 0, force: 0) from 1280x800
05:660 00:021 OCC: Requesting 1280x800@0 (max: 0) resolution, curr 0, total 1
05:682 00:021 OCC: Current FB at 0x80000000 (0x300000), format 1, res 1024x768 scan 1024
05:704 00:021 OCC: Mode 0 - 1024x768:1
05:728 00:023 OCC: No compatible mode for 1280x800@0 (max: 0) resolution
05:749 00:021 OC: Changed resolution to 1280x800@0 (max: 0, force: 0) from 1280x800 - Not Found
05:771 00:021 OCC: Using builtin text renderer with 1 scale
05:793 00:022 OCC: Install console control (753AD9C0/0/0), current - Success
05:815 00:021 OCC: Setup ASCII Output - Success
05:837 00:022 OC: Requested console mode is 0x0 (max: 0) from
05:862 00:024 OC: Requested not to use audio
05:886 00:024 OC: OcMiscLoadSystemReport...
05:908 00:022 OC: OcLoadAcpiSupport...

Changed to:
<key>Resolution</key> <string>1024x768</string>

Result log:

05:622 00:021 OCC: GOP exists on ConsoleOutHandle and has 1 modes
05:643 00:021 OC: Requested resolution is 1024x768@0 (max: 0, force: 0) from 1024x768
05:665 00:021 OCC: Requesting 1024x768@0 (max: 0) resolution, curr 0, total 1
05:687 00:021 OCC: Current FB at 0x80000000 (0x300000), format 1, res 1024x768 scan 1024
05:709 00:021 OCC: Mode 0 - 1024x768:1
05:730 00:021 OCC: Current mode matches desired mode 0
05:752 00:021 OC: Changed resolution to 1024x768@0 (max: 0, force: 0) from 1024x768 - Already started
05:774 00:021 OC: Setting UIScale to 1 - Success
05:797 00:023 OCC: Using builtin text renderer with 1 scale
05:822 00:024 OCC: Install console control (753AD9C0/0/0), current - Success
05:843 00:021 OCC: Setup ASCII Output - Success
05:865 00:022 OC: Requested console mode is 0x0 (max: 0) from
05:888 00:022 OC: Requested not to use audio
05:910 00:022 OC: OcMiscLoadSystemReport...
05:934 00:024 OC: OcLoadAcpiSupport...

Changed to:
<key>GopPassThrough</key> <string>Enabled</string>

Result log:

04:961 00:014 OCC: Found 2 handles with GOP draw
04:976 00:015 OCC: Found 2 handles with Apple Framebuffer info
04:991 00:014 OCC: AppleFramebufferInfo - Got Base 80000000, Size 3145728, RowBytes 4096, Width 1024, Height 768, Depth 32
05:006 00:015 OCC: Found 2 handles with UGA for GOP check
05:021 00:014 OCC: Trying handle 0 - 75D2F118
05:035 00:014 OCC: Failed to retrieve AppleFramebufferInfo protocol on handle 0 - 75D2F118 (Unsupported)
05:050 00:014 OCC: Trying handle 1 - 75BA6498
05:067 00:017 OCC: Skipping GOP proxying as it is already present on handle 1 - 75BA6498
05:082 00:014 OCC: GOP exists on ConsoleOutHandle and has 1 modes
05:096 00:014 OC: Requested resolution is 0x0@0 (max: 1, force: 0) from Max
05:111 00:014 OCC: Requesting 0x0@0 (max: 1) resolution, curr 0, total 1
05:125 00:014 OCC: Current FB at 0x80000000 (0x300000), format 1, res 1024x768 scan 1024
05:140 00:014 OCC: Mode 0 - 1024x768:1
05:154 00:014 OCC: Current mode matches desired mode 0
05:169 00:014 OC: Changed resolution to 0x0@0 (max: 1, force: 0) from Max - Already started
05:183 00:014 OCC: Using builtin text renderer with 1 scale
05:200 00:017 OCC: Install console control (753B39C0/0/0), current - Success
05:217 00:016 OCC: Setup ASCII Output - Success
05:234 00:017 OC: Requested console mode is 0x0 (max: 0) from
05:249 00:015 OC: Requested not to use audio
05:264 00:015 OC: OcMiscLoadSystemReport...
05:279 00:014 OC: OcLoadAcpiSupport...

Changed to:
<key>UgaPassThrough</key> <true/>

Result log:

05:489 00:019 OCC: GOP exists on ConsoleOutHandle and has 1 modes
05:510 00:020 OC: Requested resolution is 0x0@0 (max: 1, force: 0) from Max
05:532 00:022 OCC: Requesting 0x0@0 (max: 1) resolution, curr 0, total 1
05:554 00:021 OCC: Current FB at 0x80000000 (0x300000), format 1, res 1024x768 scan 1024
05:575 00:020 OCC: Mode 0 - 1024x768:1
05:596 00:020 OCC: Current mode matches desired mode 0
05:615 00:019 OC: Changed resolution to 0x0@0 (max: 1, force: 0) from Max - Already started
05:635 00:019 OCC: Found 2 handles with UGA draw
05:655 00:020 OCC: Found 2 handles with GOP for UGA check
05:676 00:020 OCC: Trying handle 0 - 75D5D298
05:699 00:022 OCC: Skipping UGA proxying as it is already present on handle 0 - 75D5D298
05:719 00:020 OCC: Trying handle 1 - 75BA2498
05:740 00:020 OCC: Skipping UGA proxying as it is already present on handle 1 - 75BA2498
05:761 00:020 OCC: Using builtin text renderer with 1 scale
05:782 00:021 OCC: Install console control (753AD9C0/0/0), current - Success
05:805 00:022 OCC: Setup ASCII Output - Success
05:826 00:021 OC: Requested console mode is 0x0 (max: 0) from
05:848 00:021 OC: Requested not to use audio
05:869 00:021 OC: OcMiscLoadSystemReport...
05:890 00:021 OC: OcLoadAcpiSupport...

Changed to:
<key>ForceResolution</key> <true/>

Result log:

06:095 00:026 OCC: GOP exists on ConsoleOutHandle and has 1 modes
06:122 00:026 OC: Requested resolution is 0x0@0 (max: 1, force: 1) from Max
06:147 00:025 OCC: Missing OcForceResolution protocol - Not Found
06:174 00:026 OCC: Requesting 0x0@0 (max: 1) resolution, curr 0, total 1
06:203 00:028 OCC: Current FB at 0x80000000 (0x300000), format 1, res 1024x768 scan 1024
06:230 00:026 OCC: Mode 0 - 1024x768:1
06:258 00:027 OCC: Current mode matches desired mode 0
06:285 00:026 OC: Changed resolution to 0x0@0 (max: 1, force: 1) from Max - Already started
06:312 00:026 OCC: Using builtin text renderer with 1 scale
06:339 00:027 OCC: Install console control (753AB9C0/0/0), current - Success
06:366 00:026 OCC: Setup ASCII Output - Success
06:394 00:027 OC: Requested console mode is 0x0 (max: 0) from 
06:421 00:027 OC: Requested not to use audio
06:451 00:029 OC: OcMiscLoadSystemReport...
06:478 00:027 OC: OcLoadAcpiSupport...

In all cases - a black screen. I can't get the OC Bootscreen to appear yet.

 

[jsts]

The display is single link DVI so any of the following connection types should work:
- HDMI
- DisplayPort dual mode (DP++) passive adaptor (to HDMI or DVI) (GPUs still support dual mode, right?)
- DisplayPort active adaptor (to HDMI or DVI)

To check (and for fun ,-) I also ordered an inexpensive HDMI to DisplayPort, BENFEI HDMI to DisplayPort Adapter Resolution Up to 4K @ 60 Hz Compatible with Laptop, Xbox 360 One, PS4 PS3 HDMI Device (HDMI Input to DisplayPort Output).

Even with all the info, I don't know if there's a way to make the RX 6600 XT choose the correct resolution mode. Or modify Open Core log to show EDID protocol handles (this would be easiest).

Am I correct in assuming that it is necessary to enable Kext debug?

 

[startergo]

In all cases - a black screen. I can't get the OC Bootscreen to appear yet.

Try this:

1. ReconnectGraphicsOnConnect
   Type: plist boolean
   Failsafe: false
   Description: Reconnect all graphics drivers during driver connection.
   On certain firmware, it may be desireable to use an alternative graphics driver, for example BiosVideo.efi, providing better screen resolution options on legacy machines, or a driver supporting ForceResolution. This option attempts to disconnect all currently connected graphics drivers before connecting newly loaded drivers.
   Note: This option requires ConnectDrivers to be enabled.

and this:

1. ForgeUefiSupport
   Type: plist boolean
   Failsafe: false
   Description: Implement partial UEFI 2.x support on EFI 1.x firmware.
   This setting allows running some software written for UEFI 2.x firmware like NVIDIA GOP Option ROMs on hardware with older EFI 1.x firmware like MacPro5,1.

 

[Dayo]

ForgeUEFI must be used along with ReloadROM for the intended effect on GPUs. This however, is not likely to solve the issues faced by the OP as it is not due to a failure to load the GOP driver. ReconnectGraphics might be worth a shot though.

Irony is that the GPU would most likely have worked without the flashing. Suggestion to OP is to make a bug report on the OpenCore BugTracker on GitHub with the outputs of Post 31. Should at the least give you a clear idea of what is going wrong.

 

[joevt]

Found EFI/OC/config.plist:

and then modified to:

<key>DisplayLevel</key> <integer>2147483714</integer>

Reboot. But I did not find any significant changes in the log (attached opencore-2022-08-07-094644.txt.zip).

Ok, I guess OpenCore probably doesn't have much more to say about this issue. Try debug version of RefindPlus. It will create its own debug log. It can chain load OpenCore for booting unsupported macOS version but mostly we are interested in getting a boot screen in EFI or any info.

It works! I attach the AGDCDiagnose output: AGDCDiagnose.txt.zip

AllRez output attached (AllRez-Apple-LED-Cinema-Display-24.txt.zip)

Ok, All the EDIDs match what you previously sent from macOS so there's no problem there.

We don't know what EDID if any is getting read by EFI. A developer might have to step in here to get this info into a log if you can't get into the EFI Shell from OpenCore or RefindPlus (requires a working GPU or a serial port)
Here's some info about EDID protocols that a GOP driver should handle:
https://edk2-docs.gitbook.io/edk-ii-uefi-driver-writer-s-guide/23_graphics_driver_design_guidelines
Maybe we can install a EFI_EDID_OVERRIDE_PROTOCOL then restart the driver? I don't know, it seems to me the driver should have read the EDID and used the default mode or the first mode it found if it can't determine the default. If the EFI driver is getting confused, then an EDID override could provide a simplified EDID...

AllRez shows that macOS accepted the Detailed Timing Descriptor version of the 1920x1200 timing but excluded the Standard Timing version of the 1920x1200 timing.

The display appears to be connected using a DisplayPort. Is this a HDMI port of the RX 6600 XT? Perhaps it implements HDMI using an active adapter? The DisplayPort source device vendor is AMD so that may be the case. But wait - it say the sink device vendor is Apple. Does that mean the display is DisplayPort? Actually, the EDID says it has a DisplayPort interface. I'm confused. Actually, the original post said Apple LED Cinema Display 24" DisplayPort ---- I'm dumb. Forget everything I said about HDMI and DVI. I guess I got confused because all the 1920x1200 displays I've ever used were DVI.

To check (and for fun ,-) I also ordered an inexpensive HDMI to DisplayPort, BENFEI HDMI to DisplayPort Adapter Resolution Up to 4K @ 60 Hz Compatible with Laptop, Xbox 360 One, PS4 PS3 HDMI Device (HDMI Input to DisplayPort Output).

You have a DisplayPort display, so forget what I said about HDMI and DVI. However, an HDMI to DisplayPort adapter like that one might be interesting. Or maybe a scaler can be added to the chain of adapters to take a 1024x768 signal and scale it to 1920x1200? But that would probably expensive and I don't know if such a thing exists. Plus the scaler would have to match the expected timing.

Am I correct in assuming that it is necessary to enable Kext debug?

No. kexts are for macOS. The EFI environment happens before that. EFI is where the changes need to be.

 

[Dayo]

Try debug version of RefindPlus. It will create its own debug log. It can chain load OpenCore for booting unsupported macOS version but mostly we are interested in getting a boot screen in EFI or any info.

I think RefindPlus will default to the same problematic GOP setup but I suppose it might be useful to see the log.
One option could be running it in "TextOnly" mode. It might also be possible to access UEFI Shell in this mode.

 

[jsts]

Try this: ReconnectGraphicsOnConnect

Black screen. OC didn't even boot up. Even when connecting via HDMI to LG TV-panel - nothing happens at all after chime, just black screen.

and this: ForgeUefiSupport

No luck. OC - black screen, macOS - Ok.

 

[jsts]

Maybe you can ask Marek if the GPU has GOP or UGA or both. I think the point of getting GPU from MacVidCard is to get a GPU with UGA driver to use with old Mac Pros?

I wrote to Marek. I hope he won't refuse to help.

RefindPlus might be useful here as you can easily set a target resolution (will only work if available though) and the debug log will tell whether you have GOP and/or UGA. It will always pick GOP if available and fallback on UGA if available when GOP is not. In either case, availability status of both is logged (Detailed for GOP ... Available modes and resolutions etc).

I'll try that.

Suggestion to OP is to make a bug report on the OpenCore BugTracker on GitHub with the outputs of Post 31. Should at the least give you a clear idea of what is going wrong.

Thank you, I will try to post a bug report.

No. kexts are for macOS. The EFI environment happens before that. EFI is where the changes need to be.

I get it. Thank you. Then I'll try to use RefindPlus, because I can't find a manual on how to enable EDID debug in OpenCore.

 

[netkas]

>Maybe you can ask Marek if the GPU has GOP or UGA or both. I think the point of getting GPU from MacVidCard is to get a GPU with UGA driver to use with old Mac Pros?

wrong, macpro 4,1/5,1 use GOP ( apple's GOP tho, not exactly uefi's GOP), it ignores UGA
macpro 3,1 uses UGA and ignores GOP

the rom provides both to ensure compatibility with 3,1 and 4,1/5,1

Only single mode is provided, the maximum supported mode reported by the card. ( as do the original apple's cards)

So trying to change the resolution in OC is pointless.

HDMI2DP dongle is the best idea so far.

 

[startergo]

Only single mode is provided, the maximum supported mode reported by the card. ( as do the original apple's cards)

So the same card reported different maximum supported modes for different displays with different ports.

3. Then I plugged the computer with the RX 6600 XT into the LG OLED55CX (4K) panel via HDMI. Apple bootscreen appeared (!), OpenCore bootscreen also appeared and in the log was this:

00:000 00:000 OC: Starting OpenCore...
<...>
05:773 00:042 OCC: GOP exists on ConsoleOutHandle and has 1 modes
05:815 00:042 OC: Requested resolution is 0x0@0 (max: 1, force: 0) from Max
05:858 00:042 OCC: Requesting 0x0@0 (max: 1) resolution, curr 0, total 1
05:900 00:042 OCC: Current FB at 0x80000000 (0x500000), format 1, res 1280x1024 scan 1280
05:943 00:042 OCC: Mode 0 - 1280x1024:1
05:985 00:042 OCC: Current mode matches desired mode 0
06:027 00:042 OC: Changed resolution to 0x0@0 (max: 1, force: 0) from Max - Already started
06:069 00:042 OCC: Using builtin text renderer with 1 scale
06:112 00:043 OCC: Install console control (728209C0/0/0), current - Success
06:154 00:041 OCC: Setup ASCII Output - Success
06:196 00:042 OC: Requested console mode is 0x0 (max: 0) from
06:239 00:042 OC: Requested not to use audio
06:281 00:042 OC: OcMiscLoadSystemReport...
06:324 00:042 OC: OcLoadAcpiSupport...
<...>

Does this mean that the maximum supported mode by the card depends on the port used? So HDMI to DP should provide the same resolution in theory?

 

[joevt]

So the same card reported different maximum supported modes for different displays with different ports.


Does this mean that the maximum supported mode by the card depends on the port used? So HDMI to DP should provide the same resolution in theory?

Ok, so it's possible for the 6600 XT driver to report different modes, but in both examples so far it chooses a mode that is not optimal.
1024x768 is chosen for the DisplayPort 1920x1200 display even though 1024x768 is not a mode in the EDID.
1280x1024 is chosen for HDMI 3840x2160 display (unknown EDID).

Next step is to get a dh -d -v > dh_d_v.txt from EFI Shell from anyone with the same card. Then we can see if it supports EDID.
Another thing we can do is get the firmware, (modified or unmodified) and see if it references any of the EDID protocols (search for 16 byte UUID). If it does, then maybe we can modify the EDID. If not, then a more complicated patch of the firmware would be required to modify the way it chooses the list of output resolutions.

 

[joevt]

My suggestion for RefindPlus or OpenCore:

For every EDID protocol (discovered, active), dump the device path (if the handle has one), dump the EDID, and maybe dump a list of all the other protocols belonging to that handle.

Before doing that, check if there's an EDID override protocol. Override it with your own protocol, or create a new override protocol if it doesn't exist.

Add some logging code to the GetEdid function of the EDID override protocol so we can learn if the GOP calls it and with what EDID the GOP is checking for override.
https://edk2-docs.gitbook.io/edk-ii...ile_templates/readme.4/a41_edidoverridec_file

Maybe you need to reinit the GOP driver to learn anything, in case the GOP driver already called GetEdid before you could add your own override protocol.

Create an array of EDID overrides in the config:
- find = what EDID to look for
- replace = what EDID to replace
- find mask = for each bit in a byte, 1 = match bit, 0 = match any bit.
- replace mask = for each bit in a byte, 1 = replace bit, 1 = don't change bit.
// Empty find = replace all EDIDs
// Empty replace = don't override anything
// Empty find mask = match all bytes
// Empty replace mask = replace all bytes

 

[Dayo]

Seems more of a task for OpenCore ... at least the overrides implementation bit.
Perhaps one of their devs might take it on if you raise a request on their tracker?

 

[jsts]

On Marek's advice I purchased a brand new mDP-DP cable. Replacing the cable did not solve the issue. The issue, as expected, is not the cables.

I also bought an active BENFEI HDMI to DisplayPort Adapter - and that doesn't solve the problem either.

OpenCore logs when connected via the BENFEI HDMI to DisplayPort Adapter:

05:086 00:020 OCC: GOP exists on ConsoleOutHandle and has 1 modes
05:105 00:018 OC: Requested resolution is 0x0@0 (max: 1, force: 0) from Max
05:124 00:018 OCC: Requesting 0x0@0 (max: 1) resolution, curr 0, total 1
05:142 00:018 OCC: Current FB at 0x80000000 (0x7E9000), format 1, res 1920x1080 scan 1920
05:161 00:018 OCC: Mode 0 - 1920x1080:1
05:179 00:018 OCC: Current mode matches desired mode 0
05:198 00:018 OC: Changed resolution to 0x0@0 (max: 1, force: 0) from Max - Already started
05:217 00:018 OCC: Using builtin text renderer with 1 scale
05:236 00:019 OCC: Install console control (753AF9C0/0/0), current - Success
05:255 00:018 OCC: Setup ASCII Output - Success
05:274 00:019 OC: Requested console mode is 0x0 (max: 0) from
05:295 00:020 OC: Requested not to use audio
05:314 00:019 OC: OcMiscLoadSystemReport...
05:333 00:019 OC: OcLoadAcpiSupport...

In general, HDMI-DP also wants to work with 1920x1080, but the monitor doesn't want it. Moreover, through the HDMI adapter, the image is not displayed at all either at the boot screen or in the operating system. There are no picture at all. Total blackscreen.

 

[Borowski]

To my knowledge, it will work only one way, from DP (card) to HDMI (display), not inverted. From HDMI to DP active adaptors are needed.

 

[joevt]

To my knowledge, it will work only one way, from DP (card) to HDMI (display), not inverted. From HDMI to DP active adaptors are needed.

What are you saying? @jsts linked a HDMI to DP active adapter.
The CAC-1335 and CAC-1336 are new HDMI to DP active adapters. They're the only ones that I know that can do > 4096 width. I was able to get 5K39 with a CAC-1336 but I had to put a Thunderbolt 4 hub between the CAC-1336 and the DisplayPort display.
https://insights.club-3d.com/thread/cac-1336-operation/

 

[jsts]

HDMI2DP dongle is the best idea so far.

Yes, it was a good idea. But it didn't work. Can you recommend anything else?

 

[startergo]

Yes, it was a good idea. But it didn't work. Can you recommend anything else?

Did you try active converters?

CAC-1335 and CAC-1336

 

[Derbyblu]

Hi jsts,
I'm having the same problem. Bought a rx 6600 from macvidcards europe for a MacPro 5.1 with 24" Cinema Display and it does not show a bootscreen.
Firmware is on 144 everything worked with bootscreen with my old radeon 4870 and catalina, but it's not metal-able.
I also tried everything you did...but no success. I also used both cards at the same time, with two different displays. The cinema display connected to the old 4870 via miniDP and the rx6600 connected via HDMItoHDMI to the newer 24" Acer display. (I created an adapter for the two superdrives and used the power from there for the rx6600 - sure not to be done for daily business but should work just for booting experience).
It showed boot screen when both cards are installed but then stopped working when osx got loaded. Windows worked fine with both cards installed.
Whatever, not the solution for me as when you give some load on it...you can imagine it powers down. I wated a card that shows bootscreen natively as advertised on MVCeu as I also have a radeon VII running with no support and no boot screen (wanted to get rid of that).

When pulling the old card, nvram reset..adjusting the nvram again and starting with only the rx6600 installed...no EFI boot screen shows up.
Only the open core bootscreen works with the stick, but not natively.
Any news on this?

BR
Chris

 

[startergo]

I would ask for a refund if I were you. Now you can buy unmodified metal 3 card and even see the boot screen with OC’s help. You may still need to apply the free syncretic vbios patch.