Ryzen 9000

[Ejh4286]

I installed a Ryzen 9700 into a Asus Strix B650E-
F (latest BIOS) that was successfully booting Sequoia Beta 4. Using latest kexts, patches, etc. I was not really expecting it to boot on the first try and I was right. See screen shot attached. Has anyone else tried to Hackintosh a Ryzen 9000 series yet?

 

[ssuper2k]

Sure, you can use Handbrake 1.8.2 with those settings on a 4K HDR/DoVI HEVC "BIG" file :

1. x265 10-bit encoder
2. multi-pass with turbo first pass
3. bitrate 30720 kbps
4. Options:
   1. presets : veryslow
   2. profile : auto
   3. level : 5.1
   4. additional options : repeat-headers=1:hdr=1:hdr-opt=1:no-dhdr10-opt=1:aud=1:ref=4:gop-lookahead=0:rc-lookahead=60:lookahead-slices=1:rect=1:amp=1:tu-inter-depth=4:tu-intra-depth=4:limit-tu=4sy-rd=2.00sy-rdoq=1.00:rdoq-level=2:dynamic-rd=0.00:nr-intra=0:nr-inter=0:max-merge=5:limit-refs=2:no-limit-modes=1:me=3:subme=7:merange=57:weightp=1:weightb=1:deblock=-3,-3:rd=4:selective-sao=4:no-fast-intra=1:b-intra=1:aq-mode=4:scenecut-aware-qp=0:bframes=8:min-keyint=23:keyint=250:scenecut=40:hist-scenecut=1:early-skip=0:rskip=2:rskip-edge-threshold=5:vbv-maxrate=160000:vbv-bufsize=160000:vbv-init=0.9:min-vbv-fullness=50.0:max-vbv-fullness=80.0:aq-strength=0.80:limit-sao=1:high-tier=1

FYI, everything works perfectly well under Windows 11 and Ubuntu 24.10 daily
Likewise, these settings worked perfectly with my 7950X and my 5950X under Sonoma.

No problem running Cinebench 2024 (quite satisfied @145TDP (PBO 195/140/200), only 72°C):
View attachment 15179

But system goes into kernel panic after 30-45 minutes under heavy loads. (before someone ask, same results CPU stock (PBO disabled,...)).

Not sure how to put all those settings, just started this a few minutes ago

Will leave it running 1h .. so far, all fans are rumbling

Edit: It worked, no problem at all

 

[Alex4querty]

I don't understand. Three different caclulations and still the wrong core count. Can you test something for me @Alex4querty? I don't think it'll boot but try disable the ProvideCurrentCpuInfo quirk and send the log. I'm wondering if this is the cause.

providecurrentcpuinfo doesnt seem to be problem here, i'm starting to think this is a bigger problem than expected and that there is little hope that it would get fixed

 

[Shaneee]

providecurrentcpuinfo doesnt seem to be problem here, i'm starting to think this is a bigger problem than expected and that there is little hope that it would get fixed

Ok leave that quirk enabled. The issue doesn't seem to be related to the AMD code within OpenCore but rather the MP services. I've adjusted it in this version and reverted the AMD code back to what I had originally.

 

[ssuper2k]

Sure, you can use Handbrake 1.8.2 with those settings on a 4K HDR/DoVI HEVC "BIG" file :

1. x265 10-bit encoder
2. multi-pass with turbo first pass
3. bitrate 30720 kbps
4. Options:
   1. presets : veryslow
   2. profile : auto
   3. level : 5.1
   4. additional options : repeat-headers=1:hdr=1:hdr-opt=1:no-dhdr10-opt=1:aud=1:ref=4:gop-lookahead=0:rc-lookahead=60:lookahead-slices=1:rect=1:amp=1:tu-inter-depth=4:tu-intra-depth=4:limit-tu=4sy-rd=2.00sy-rdoq=1.00:rdoq-level=2:dynamic-rd=0.00:nr-intra=0:nr-inter=0:max-merge=5:limit-refs=2:no-limit-modes=1:me=3:subme=7:merange=57:weightp=1:weightb=1:deblock=-3,-3:rd=4:selective-sao=4:no-fast-intra=1:b-intra=1:aq-mode=4:scenecut-aware-qp=0:bframes=8:min-keyint=23:keyint=250:scenecut=40:hist-scenecut=1:early-skip=0:rskip=2:rskip-edge-threshold=5:vbv-maxrate=160000:vbv-bufsize=160000:vbv-init=0.9:min-vbv-fullness=50.0:max-vbv-fullness=80.0:aq-strength=0.80:limit-sao=1:high-tier=1

FYI, everything works perfectly well under Windows 11 and Ubuntu 24.10 daily
Likewise, these settings worked perfectly with my 7950X and my 5950X under Sonoma.

No problem running Cinebench 2024 (quite satisfied @145TDP (PBO 195/140/200), only 72°C):
View attachment 15179

But system goes into kernel panic after 30-45 minutes under heavy loads. (before someone ask, same results CPU stock (PBO disabled,...)).

My .EFI files are signed (for my secure boot)
You can replace them with the .efi0 files (not signed)

Also fill those 00000000000000 values.

In my system, it works flawlessly, even sleep works with Sequoia

ASUS CROSSHAIR X670E HERO

 

[fabiosun]

This EFI remember me something @ssuper2k

Edit:
USB mapping used is valid for our Asus Hero, also MMIO could be different for Pro art board
Then we use kext and block useful for broadcom (and OCLP)

 

[Arch30n]

My .EFI files are signed (for my secure boot)
You can replace them with the .efi0 files (not signed)

Also fill those 00000000000000 values.

In my system, it works flawlessly, even sleep works with Sequoia

Thanks, I made 2, 3 modifications to my EFI based on yours (+ removal of AMDRyzenCPUPowerManagement and added of AMFIPass) and it's been running for over an hour now another 8 hours and we'll see, fingers crossed !

 

[sunfish]

waiting.....

 

[Ejh4286]

Ok leave that quirk enabled. The issue doesn't seem to be related to the AMD code within OpenCore but rather the MP services. I've adjusted it in this version and reverted the AMD code back to what I had originally.

I gave this a try on my Asus ROG B650E-F, BIOS 2613. Same result. I have attached my EFI and debug output.

 

[Shaneee]

I gave this a try on my Asus ROG B650E-F, BIOS 2613. Same result. I have attached my EFI and debug output.

Thanks for testing. Please try this. Based on the report earlier from Linux it shows that 14 cores are reported in the EBX register. This leads me to believe it's a firmware issue with the CPUs. This version forces 12 cores only when 14 cores are detected. Let's hope it works. Make sure you've got the correct core count set in your Kernel Patches.

 

[Arch30n]

Steady as a Rock Baby ! Many thanks to @ssuper2k, @fabiosun and corpghost for your help

I attach my EFI for those who have a ASUS ProArt X670E CREATOR WIFI.
I used the OCLP root patch + Broadcom spoof trick to make AirportItlwm.kext work on Sequoia with my AX210.
I still need to make the Thunderbolt/USB4 and the Aquantia card work, I haven't had time to go further for now. You'll have to redo the USBMap accordingly.
What doesn't work: ALC1220 for now, but this is normal under Sequoia (some use VoodooHDA instead).

 

[Ejh4286]

Thanks for testing. Please try this. Based on the report earlier from Linux it shows that 14 cores are reported in the EBX register. This leads me to believe it's a firmware issue with the CPUs. This version forces 12 cores only when 14 cores are detected. Let's hope it works. Make sure you've got the correct core count set in your Kernel Patches.

Same result. Thank you for your work. I'll keep testing as long as you're willing to work towards a solution.

 

[Shaneee]

Did it not boot at all? Did you have a previous working system with the same motherboard at all? Do you have the MMIO setup?

The OC log is going to keep reporting 14 cores regardless I think. In that last test I had 12 cores hardcoded,

          // Apply a fix for Ryzen 9 9900X which incorrectly reports 14 cores
          if (Cpu->CoreCount == 14 && Cpu->Family == 0x1A && Cpu->Model == 0x90) {
            Cpu->CoreCount = 12;  // Correct the core count for Ryzen 9 9900X
          }

Assuming this took affect which is should do, then there's another reason that the system isn't booting. Could you please dump a SysReport with OpenCore. The fact that the CPUID is reporting 14 cores is strange. A SSDT or ACPI patch may be the answer here to correct the core count.

Edit: Here's a more in depth retrieval of the cores/threads from the CPUID. Please test and upload the log.

 

[Ejh4286]

Did it not boot at all? Did you have a previous working system with the same motherboard at all? Do you have the MMIO setup?

The OC log is going to keep reporting 14 cores regardless I think. In that last test I had 12 cores hardcoded,

          // Apply a fix for Ryzen 9 9900X which incorrectly reports 14 cores
          if (Cpu->CoreCount == 14 && Cpu->Family == 0x1A && Cpu->Model == 0x90) {
            Cpu->CoreCount = 12;  // Correct the core count for Ryzen 9 9900X
          }

Assuming this took affect which is should do, then there's another reason that the system isn't booting. Could you please dump a SysReport with OpenCore. The fact that the CPUID is reporting 14 cores is strange. A SSDT or ACPI patch may be the answer here to correct the core count.

Edit: Here's a more in depth retrieval of the cores/threads from the CPUID. Please test and upload the log.

Here is the debug log and SysReport.

 

[Shaneee]

@Ejh4286 is that with the latest build? I'm not seeing my debug logs there...

 

[Ejh4286]

@Ejh4286 is that with the latest build? I'm not seeing my debug logs there...

Yes, from the attachment on post 192 of this thread.

 

[Shaneee]

It's almost as if the 9900X doesn't provide the core count through the CPUID 0x8000001E. I've added some debug statements outwith this check to confirm and added a fallback. Please test and report with the OC log. Thanks.

 

[ssuper2k]

My .EFI files are signed (for my secure boot)
You can replace them with the .efi0 files (not signed)

Also fill those 00000000000000 values.

In my system, it works flawlessly, even sleep works with Sequoia

ASUS CROSSHAIR X670E HERO

Updated to Opencore Release 1.02 (final, not beta)

NO changes needed, Zero Problems

Releases · acidanthera/OpenCorePkg

OpenCore bootloader. Contribute to acidanthera/OpenCorePkg development by creating an account on GitHub.

github.com

 

[Shaneee]

@Ejh4286 have you tested latest one I uploaded?

 

[Ejh4286]

It's almost as if the 9900X doesn't provide the core count through the CPUID 0x8000001E. I've added some debug statements outwith this check to confirm and added a fallback. Please test and report with the OC log. Thanks.

Here ya go.

 

[Shaneee]

Here ya go.

Strange, nothing in the log I'm looking for. Can you check DisplayLevel in your config. Set it to 2147483647. This will all debug logs are captured.