Ryzen 7000 Testing

[ExtremeXT]

I am entertaining myself with the Ryzentosh... I never thought that the audio would end up working...

And everything works 100%, even being on the same network as the laptop, the ryzen has connected with universal control

View attachment 8590

Works just as well as Alder Lake and Raptor Lake

And all thanks to you who are geniuses .. good job

MacPro7,1 - Geekbench

Benchmark results for a MacPro7,1 with an AMD Ryzen 9 7950X processor.

browser.geekbench.com

MacPro7,1 - Geekbench

Benchmark results for a MacPro7,1 with an AMD Ryzen 9 7950X processor.

browser.geekbench.com

View attachment 8591

Honestly, it probably works even better than Alder/Raptor Lake as our Ryzen CPUs don't have the new core topology.

 

[PoMpIs]

Honestly, it probably works even better than Alder/Raptor Lake as our Ryzen CPUs don't have the new core topology.

It can be, although the Alder Lake and Raptor Lake in MacOS if it is not used more than 60% of the CPU always uses the P-Core, the e-core uses them in renders and hard work

Check out the 13900k at 5.3Ghz P-Core and 4.3Ghz E-core:

MacPro7,1 - Geekbench

Benchmark results for a MacPro7,1 with an Intel Core i9-13900K processor.

browser.geekbench.com

 

[svan71]

Is there a board that is more recommended than others?

 

[CaseySJ]

I am entertaining myself with the Ryzentosh... I never thought that the audio would end up working...

And everything works 100%, even being on the same network as the laptop, the ryzen has connected with universal control

Works just as well as Alder Lake and Raptor Lake

And all thanks to you who are geniuses .. good job

MacPro7,1 - Geekbench

Benchmark results for a MacPro7,1 with an AMD Ryzen 9 7950X processor.

browser.geekbench.com

MacPro7,1 - Geekbench

Benchmark results for a MacPro7,1 with an AMD Ryzen 9 7950X processor.

browser.geekbench.com

I thought you had returned the Ryzen 7 processor and AM5 motherboard?

If not, it's good to you back on the bandwagon!

 

[CaseySJ]

Is there a board that is more recommended than others?

Although it's too early to make a single recommendation from a wide field in which there are X670 and X670E boards from several manufacturers, I can make a couple of early remarks:

• Avoid Gigabyte X670 and X670E boards due to BIOS (DSDT) problems. Poor @ManuelB has been struggling with his Gigabyte X670E AORUS Master.
• Asus X670/X670E boards are the ones I can personally recommend at this time
• As users adopt other boards (ASRock, MSI, etc.) we'll get a better picture

 

[PoMpIs]

I thought you had returned the Ryzen 7 processor and AM5 motherboard?

If not, it's good to you back on the bandwagon!

I have returned the motherboard, now I have an Asus x670E Hero, much better to adjust the processor well, which I really like to pass benchmarks on windows..

And I have also bought UNRAID for a long time and I have fun trying Hackinabox, and virtual machines with PCI passthrougth, it is very fun to have MacOS and windows running at the same time on a single computer

 

[CaseySJ]

...
And everything works 100%, even being on the same network as the laptop, the ryzen has connected with universal control
...

Speaking of Universal Control, I was connecting both my Intel and Ryzen systems to a single 27" 4K monitor until a couple weeks ago when I purchased a 2nd 27" 4K monitor. I was using a 4-port USB KVM for switching my keyboard and mouse between the two systems. But last night as I was testing the revised OpenCore.efi on both Ryzen and Intel systems, I had both computers powered on at the same time and both monitors powered on at the same time (which is rare for me).

Foolishly thinking that both monitors were connected to the same computer, I instinctively dragged the OpenCore.efi file from one monitor all the way across to the other monitor. And the mouse -- and the file -- seamlessly moved to the other monitor. At that moment I suddenly realized what had happened -- the mouse and the file had crossed over from the Ryzen system to the Intel system effortlessly!

I was stunned, as this was my first real experience with Universal Control. I had tested it once before under "lab" conditions, and never used it since. But that changed last night.

My dual system arrangement. Universal Control bridges the two systems quite seamlessly.

 

[Shaneee]

@Shaneee,

Because the credit for this belongs to you, we would like you to consider submitting the Pull Request to OpenCorePkg repo on GitHub.

The final code is as follows:

  //
  // Perform TSC and FSB calculations. This is traditionally done in tsc.c in XNU.
  //
  // For AMD Processors
  if (CpuInfo->Family == 0xF && (CpuInfo->ExtFamily == 0x8 || CpuInfo->ExtFamily == 0xA)) {
        DEBUG ((DEBUG_INFO, "OCAK: Setting FSB and TSC for Family 0x%x and ExtFamily 0x%x\n", (UINT16)CpuInfo->Family, (UINT16)CpuInfo->ExtFamily));
        busFreqValue    = CpuInfo->FSBFrequency;
        busFCvtt2nValue = DivU64x64Remainder ((1000000000ULL << 32), busFreqValue, NULL);
        busFCvtn2tValue = DivU64x64Remainder(((1000000000ULL) << 32), busFCvtt2nValue, NULL);

        tscFreqValue    = CpuInfo->CPUFrequency;
        tscFCvtt2nValue = DivU64x64Remainder ((1000000000ULL << 32), tscFreqValue, NULL);
        tscFCvtn2tValue = DivU64x64Remainder(((1000000000ULL)  << 32), tscFCvtt2nValue, NULL);
  }
  // For all other processors
  else {
        busFreqValue    = CpuInfo->FSBFrequency;
        busFCvtt2nValue = DivU64x64Remainder ((1000000000ULL << 32), busFreqValue, NULL);
        busFCvtn2tValue = DivU64x64Remainder (0xFFFFFFFFFFFFFFFFULL, busFCvtt2nValue, NULL);

        tscFreqValue    = CpuInfo->CPUFrequency;
        tscFCvtt2nValue = DivU64x64Remainder ((1000000000ULL << 32), tscFreqValue, NULL);
        tscFCvtn2tValue = DivU64x64Remainder (0xFFFFFFFFFFFFFFFFULL, tscFCvtt2nValue, NULL);
  }

I'll get right on it DhinakG says there's a better chance of it being accepted if it's explained well and clean. I've already formatted it to match the rest of the file.
I was thinking of the explanation,

Alter TSC and Bus calculations for AMD CPUs with integrated graphics to fix crackling audio issue.

Should probably change the // For AMD Processors to mention APUs as that's what it's applying to correct?

 

[CaseySJ]

I'll get right on it DhinakG says there's a better chance of it being accepted if it's explained well and clean. I've already formatted it to match the rest of the file.
I was thinking of the explanation,

Should probably change the // For AMD Processors to mention APUs as that's what it's applying to correct?

Although the TSC/FSB fix is for AMD processors that suffer from audio stutter, we decided to apply the change to all AMD processors whether or not they have an on-board GPU. We have tested the fix on both AMD APUs and CPUs across both AM4 and AM5 platforms, and found every system to behave correctly.

 

[PoMpIs]

@CaseySJ

If the universal control is a great invention... I have several computers, and when they updated to beta 11 of Ventura I had them turned on and connected to the same network and I could control them all with the Alder Lake magic trackpad...

it works like a charm..

And it is also very good to be able to use the iPhone as a camera, click here:



and the camera is activated on the iPhone

 

[Shaneee]

Although the TSC/FSB fix is for AMD processors that suffer from audio stutter, we decided to apply the change to all AMD processors whether or not they have an on-board GPU. We have tested the fix on both AMD APUs and CPUs across both AM4 and AM5 platforms, and found every system to behave correctly.

Great Just need a decent explanation of it then to keep them happy lol Currently I have it titled, TSC/FSB Fix for AMD CPUs.

 

[CaseySJ]

Great Just need a decent explanation of it then to keep them happy lol Currently I have it titled, TSC/FSB Fix for AMD CPUs.

Perhaps something like this:

Initial TSC frequency calculation for AMD processors differs somewhat from the frequency calculation for Intel processors. Without special treatment for AMD processors, macOS suffers from TSC de-sync particularly if the AMD processor has on-board GPU. This in turn leads to significant amounts of audio stutter across the entire audio subsystem. This PR has been tested on both AMD and Intel processors to ensure proper operation for both platforms.

Everyone is welcome to edit/refine this explanation.

 

[Shaneee]

Fingers crossed, https://github.com/acidanthera/OpenCorePkg/pull/396

Edit: Failed the uncrustify test. Seems it's not happy with the codestyle.

 

[CaseySJ]

Interim Update #17:

• We have completed the modification of OpenCore to support FSB/TSC frequency calculations for AMD processors
• A pull request has been issued; if accepted in time, the fix will appear in OpenCore 0.8.7 on the first Monday of December
• @ExtremeXT and I are helping @ManuelB with post-installation problems on Gigabyte X670E AORUS Master
  • This board has some BIOS (DSDT) problems that require one or more ACPI patches
  • User is experiencing issues with sleep, wake, Intel Bluetooth and Intel WiFi
  • We've gone through an exhaustive USB port mapping procedure, which may be helpful to those facing this task
• As people begin to ask where is the build guide for AM5 we may need to turn our attention to that topic next
• We are also starting to look into Aquantia 10GbE connectivity issues on Monterey and Ventura
  • There may not be an easy answer for this, so we should temper our expectations
  • I've installed the card in my X670E system, and the driver recognizes and attaches to the card in both Monterey and Ventura, but when an Ethernet cable is connected, both operating systems freeze
  • This is emblematic of Vt-d issues

 

[CaseySJ]

...

Edit: Failed the uncrustify test. Seems it's not happy with the codestyle.

I believe Uncrustify wants it like this:

  //
  // Perform TSC and FSB calculations. This is traditionally done in tsc.c in XNU.
  //
  // For AMD Processors
  if ((CpuInfo->Family == 0xF) && ((CpuInfo->ExtFamily == 0x8) || (CpuInfo->ExtFamily == 0xA))) {
    DEBUG ((DEBUG_INFO, "OCAK: Setting FSB and TSC for Family 0x%x and ExtFamily 0x%x\n", (UINT16)CpuInfo->Family, (UINT16)CpuInfo->ExtFamily));
    busFreqValue    = CpuInfo->FSBFrequency;
    busFCvtt2nValue = DivU64x64Remainder ((1000000000ULL << 32), busFreqValue, NULL);
    busFCvtn2tValue = DivU64x64Remainder ((1000000000ULL << 32), busFCvtt2nValue, NULL);

    tscFreqValue    = CpuInfo->CPUFrequency;
    tscFCvtt2nValue = DivU64x64Remainder ((1000000000ULL << 32), tscFreqValue, NULL);
    tscFCvtn2tValue = DivU64x64Remainder ((1000000000ULL  << 32), tscFCvtt2nValue, NULL);
  }
  // For all other processors
  else {
    busFreqValue    = CpuInfo->FSBFrequency;
    busFCvtt2nValue = DivU64x64Remainder ((1000000000ULL << 32), busFreqValue, NULL);
    busFCvtn2tValue = DivU64x64Remainder (0xFFFFFFFFFFFFFFFFULL, busFCvtt2nValue, NULL);

    tscFreqValue    = CpuInfo->CPUFrequency;
    tscFCvtt2nValue = DivU64x64Remainder ((1000000000ULL << 32), tscFreqValue, NULL);
    tscFCvtn2tValue = DivU64x64Remainder (0xFFFFFFFFFFFFFFFFULL, tscFCvtt2nValue, NULL);
  }

 

[Shaneee]

UncrustifyThe whole damn file seems to be off lol

Edit: Downloaded the artifact diff and it is indeed just our modification Let me commit the changes you've shared there with an Uncrustify fix tag.

 

[CaseySJ]

UncrustifyThe whole damn file seems to be off lol

Edit: Downloaded the artifact diff and it is indeed just our modification Let me commit the changes you've shared there with an Uncrustify fix tag.

Yes, just these lines that I fixed above:

 

[Shaneee]

Ohh come on I made it worse

 

[CaseySJ]

@Shaneee

Ohh come on I made it worse

No worries -- Uncrustify actually gives us the right answer. Please see this by clicking here:

TSC/FSB Fix for AMD CPUs · acidanthera/OpenCorePkg@53c44a3

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

github.com

 

[CaseySJ]

@Shaneee

We're almost there...it wants everything left-shifted by 2 spaces.

Yes I hate that thing too.

It took me 10 or more attempts to get past the Lint code-style checker when I submitted the liquidctl driver for Asus AURA LED controller.