Contribute
Register

GA-H97-D3H kernel panic on waking from deep sleep

Status
Not open for further replies.
Joined
May 13, 2014
Messages
163
Motherboard
Gigabyte Z390 Designare
CPU
i9-9900K
Graphics
Radeon VII
Hi guys, I'm an on-and-off Hackintosher from back in the Leopard 10.5 days, so not really a novice, but had a bit of hiatus (I owned a real Mac Pro 5.1 and now I have a Hackintosh) so I'm currently re-learning.

I've had Mavericks installed previously (over a period of around 6 months) without much issue. Originally installed with Unibeast and post-install tweaks done with Multibeast (default DSDT-free settings).

For Yosemite, I decided to try out Clover instead so I followed the TonyMacX86guide and the install went really well (minus ethernet, which I fixed with Intel1000e kext) and it worked pretty much 'out of the box'.

The only things I've had problems with are:

  • Waking from 'deep sleep' (ie. After sleeping longer than an hour or so)
  • Turbo Boost

Specs:
i7-4790
Gigabyte H97-D3H
GTX780
Coolermaster Vanguard V750 PSU (note: doesn't support Haswell C6/C7 states)
16GB RAM
1x250GB Samsung EVO SSD (OS X Yosemite 10.10)
1x480GB Crucial M500 SSD (Windows 8.1)
Broadcom BCM94360CD Bluetooth 4.0/WiFi AC card

SMBIOS: iMac 14,2

Kexts currently installed in EFI/Clover/kexts:

  • FakeSMC.kext
  • AppleIntel1000e.kext
  • GenericUSBXHCI.kext

UEFI settings:

  • Load optimised defaults
  • Advanced memory settings > XMP Profile 1
  • VT-d disabled
  • XHCI mode Auto
  • Intel Processor Graphics Disabled
  • XHCI Hand-off Enabled
  • EHCI Hand-off Enabled
  • SATA configuration AHCI

Current boot flags:
-v dart=0 darkwake=0 nvda_drv=1

Any suggestions? Waking from deep sleep currently results in a kernel panic like the attached log.
View attachment Kernel Panic - 12 11 2014.txt

To spot Turbo Boost not going up to 4GHz, I've monitored clocks with Intel Power Gadget while the system is under stress tests.

EDIT: Sorry I just realised someone else posted about the M version of this motherboard while I was in the middle of typing this all out, although they didn't provide much system detail.

Also just realised there's others:
http://www.tonymacx86.com/general-help/148289-memory-corrupted-after-longer-sleep.html
 
I posted an update in the other thread about my unsuccessful testing using a SSDT, but I wanted to note that I'm pretty sure your PSU is Haswell certified. Cooler Master has had some inconsistent and out of date info posted on their site about this, but I checked with them about the V550 PSU I bought and they confirmed it is Haswell certified. I now see they list both our models as Haswell certified on these pages:
http://us.coolermaster.com/product/Detail/powersupply/vsm-psu/v550.html
http://us.coolermaster.com/product/Detail/powersupply/vsm-psu/v750.html
Some of their other pages still don't have this information, though. I just wanted to mention it in case you were worried that your PSU might be the cause of this problem.
 
I posted an update in the other thread about my unsuccessful testing using a SSDT, but I wanted to note that I'm pretty sure your PSU is Haswell certified. Cooler Master has had some inconsistent and out of date info posted on their site about this, but I checked with them about the V550 PSU I bought and they confirmed it is Haswell certified. I now see they list both our models as Haswell certified on these pages:
http://us.coolermaster.com/product/Detail/powersupply/vsm-psu/v550.html
http://us.coolermaster.com/product/Detail/powersupply/vsm-psu/v750.html
Some of their other pages still don't have this information, though. I just wanted to mention it in case you were worried that your PSU might be the cause of this problem.

Ah okay that's great to hear! Thanks for the info. When I was researching which CPU to buy, it was confusing because their media release on the site never mentioned their PSUs in the x50 range.

For now I'm just getting around the 'no sleep' thing by not sleeping and just shutting down more often. SpeedStep works fine (confirmed by Intel Power Gadget) so that's something. Hopefully there's something we can do about getting full sleep functionality in future.
 
Hey guys, FYI I've been trying out iMac15,1 BIOS and it seems to be going fine so far without any kernel panic from wake. Hopefully stays that way from now on!
 
Hi guys, I'm an on-and-off Hackintosher from back in the Leopard 10.5 days, so not really a novice, but had a bit of hiatus (I owned a real Mac Pro 5.1 and now I have a Hackintosh) so I'm currently re-learning.

I've had Mavericks installed previously (over a period of around 6 months) without much issue. Originally installed with Unibeast and post-install tweaks done with Multibeast (default DSDT-free settings).

For Yosemite, I decided to try out Clover instead so I followed the TonyMacX86guide and the install went really well (minus ethernet, which I fixed with Intel1000e kext) and it worked pretty much 'out of the box'.

The only things I've had problems with are:

  • Waking from 'deep sleep' (ie. After sleeping longer than an hour or so)
  • Turbo Boost

Specs:
i7-4790
Gigabyte H97-D3H
GTX780
Coolermaster Vanguard V750 PSU (note: doesn't support Haswell C6/C7 states)
16GB RAM
1x250GB Samsung EVO SSD (OS X Yosemite 10.10)
1x480GB Crucial M500 SSD (Windows 8.1)
Broadcom BCM94360CD Bluetooth 4.0/WiFi AC card

SMBIOS: iMac 14,2

Kexts currently installed in EFI/Clover/kexts:

  • FakeSMC.kext
  • AppleIntel1000e.kext
  • GenericUSBXHCI.kext

UEFI settings:

  • Load optimised defaults
  • Advanced memory settings > XMP Profile 1
  • VT-d disabled
  • XHCI mode Auto
  • Intel Processor Graphics Disabled
  • XHCI Hand-off Enabled
  • EHCI Hand-off Enabled
  • SATA configuration AHCI

Current boot flags:
-v dart=0 darkwake=0 nvda_drv=1

Any suggestions? Waking from deep sleep currently results in a kernel panic like the attached log.
View attachment 113279

To spot Turbo Boost not going up to 4GHz, I've monitored clocks with Intel Power Gadget while the system is under stress tests.

EDIT: Sorry I just realised someone else posted about the M version of this motherboard while I was in the middle of typing this all out, although they didn't provide much system detail.

Also just realised there's others:
http://www.tonymacx86.com/general-help/148289-memory-corrupted-after-longer-sleep.html

Hi,

I think deep sleep can be patched via enabling power management (don't quote me on it), and Turbo Boost can ONLY be enabled with proper power management. There's instructions somewhere on this forum on how to do that...

Cheers!
 
Hi,

I think deep sleep can be patched via enabling power management (don't quote me on it), and Turbo Boost can ONLY be enabled with proper power management. There's instructions somewhere on this forum on how to do that...

Cheers!

Power management is working natively so far and I've confirmed Turbo Boost is working as well via Intel Power Gadget; I ran Geekbench and saw my CPU ramp up to 4GHz.

Geekbench 32-bit: http://browser.primatelabs.com/geekbench3/1336631
Geekbench 64-bit: http://browser.primatelabs.com/geekbench3/1336619

Not sure why RAM shows as 0MHz in Geekbench, but no problems otherwise.
 
Just an update for those who might have the same problem.

I ended up having to patch AppleGPUPowerManagement (AGPM) to get rid of this error that I figured was stopping complete system power management from functioning:

28/11/2014 9:26:47.000 am kernel[0]: [AGPM Controller] unknownPlatform

I followed these instructions from the IM forums:
Just to share some findings, I have been trying to sort out my GTX 780 power management.

I've got the card running OOB using native Apple drivers, and it works well with nVidia web drivers too. I'm currently using nvidia web driver 343.01.01f01. My card is GK110, with device id 1004.

I have seen from other forum post saying that GPU power management depends on SMBios.plist used, and whether your DSDT table says IGPU or GFX0. For example, I have used MacBookPro11,2 SMBios.plist to get extra p-states for my CPU, but then AGPM Controller will say unknown device in the log as MacbookPro is using IGPU for matching instead of GFX0.

So I changed my SMBios.plist back to iMac14,2 and I got AGPM kext loaded but no power management was matched properly to my GFX0. I changed the Info.plist in AppleGraphicsPowerManagement.kext to match my device id. I have chosen Vendor10deDevice119d which I think is GTX 780M and changed it to my device id - Vendor10deDevice1004.

On reboot I can see that AGPM controller is loaded properly - kernel[0]: [AGPM Controller] build GPUDict by Vendor10deDevice1004

I then run some benchmark test which showed massive difference, without the patch the Cinebench R15 OpenGL test only managed to get around 60fps, and with the patch it runs at 105 fps.

Now it reads:
28/11/2014 10:44:59.000 am kernel[0]: [AGPM Controller] build GPUDict by Vendor10deDevice1004

I did some benchmark testing and my CineBench results increased significantly as a result.

cinebench-before.png


cinebench-after.png


In combination with all my C States and P States being picked up properly, I've now got fully functioning sleep.
 
Hi skittlebrau,

Glad to hear that you have fixed your problem (at least I hope it still works)

I'm trying to fix what appears to be the same issue. However, switching to iMac15,1 didn't help me at all, the only change being that I've had to fix AGPUPM. The panics are still there after longer sleep.

Can you please share your SMBIOS settings (excluding serial numbers of course)? I've seen several variants of 15,1 settings floating around, maybe the devil is in the details...

Thanks,
Stefan
 
Status
Not open for further replies.
Back
Top