Contribute
Register

Slow Boot time with Samsung 960 Evo NVME?

Status
Not open for further replies.
Hi RehabMan, thanks for the response, I really appreciate your hints. Just to be sure I'm not misunderstanding your proposals:

>> - FixUSB_1000 not needed. remove.

* Removed.

>> - CPU PM not implemented

I have no idea of what should I do so I just tried to apply what I found from posts of other similar motherboards/cpus:

* Compiled and copied /Volumes/EFI/EFI/Clover/ACPI/patched/SSDT-PluginType1.am
* AppleIntelCPUPM => true
* KernelPM => true
* Generate C States => true
* Generate P States => true

Please correct me if this is completely wrong, I would appreciate a link to the right direction.

>> - native support for SATA 8086:a282 is missing (add SATA-200-series-unsupported.kext)

* I couldn't find the "official" source, but the kext I found is just an Info.plist which seems to inject 8086:a282 to com.apple.driver.AppleAHCIPort. And it loads fine.

>> - kexts not installed correctly (all kexts should be installed to the system volume)

* Copied all kexts to /System/Library/Extensions
* Flushed kext cache, permissions, etc...

>> - USB not implemented correctly (see my USBInjectAll guide)

* Added patch "Change 15 port limit to 24 in XHCI kext 10.13" to AppleUSBXHCIPCI under Clover's KextsToPatch
* Renamed EHC1->EH01
* Renamed EHC2->EH02
* Renamed XHCI->XHC
* Renamed XHC1->XHC

That fixed USB 3.0 devices previously were not working (only when plugging them on USB 2.0 socket)

>> - VirtualBox tends to interfere with USB. remove.
>> - you have a bunch of things hooking your USB (Google Chrome, Plex Tuner Service)

That's crap but I guess I will have to live with it?

>> - check boot logs for other things that may cause delays

Unfortunately none of the above fixed the slow boot time, but I could see the huge delay was caused during the lines (see attached capture):

`fusion_wbc_thread_shutdown:1130: Consider Fusion WBC Elevator thread of container XXX`

However and after all, I managed to fix the huge slowdown, I still don't know the real issue (maybe a faulty hfs+ conversion to apfs on an earlier beta) but the following fixed it:

My primary boot disk had two partitions one with 10.12.6 (first partition, HFS+) and another with 10.13 GM (second partition, APFS).

* Deleted the 10.12.6 partition (HFS+)
* Created a new APFS container with a new APFS partition using the free space from the first deleted partition
* Used CCC to copy the 10.13 GM partition to the new APFS partition
* Deleted the old 10.13 GM partition
* Resized the first partition to fit the whole disk.

The boot is now as blazing fast as when I was on 10.12.6 with HFS+.

Thank you so much for your support :)
 

Attachments

  • IMG_20170922_152927.jpg
    IMG_20170922_152927.jpg
    4.7 MB · Views: 182
I have no idea of what should I do so I just tried to apply what I found from posts of other similar motherboards/cpus:

* Compiled and copied /Volumes/EFI/EFI/Clover/ACPI/patched/SSDT-PluginType1.am
* AppleIntelCPUPM => true
* KernelPM => true
* Generate C States => true
* Generate P States => true

Using SSDT-PluginType1.aml is fine.
But config.plist/ACPI/SSDT/Generate should be set false.

* I couldn't find the "official" source, but the kext I found is just an Info.plist which seems to inject 8086:a282 to com.apple.driver.AppleAHCIPort. And it loads fine.

The SATA 200-series kext is linked from my Clover guide:
https://www.tonymacx86.com/threads/guide-booting-the-os-x-installer-on-laptops-with-clover.148093/

>> - USB not implemented correctly (see my USBInjectAll guide)

* Added patch "Change 15 port limit to 24 in XHCI kext 10.13" to AppleUSBXHCIPCI under Clover's KextsToPatch
* Renamed EHC1->EH01
* Renamed EHC2->EH02
* Renamed XHCI->XHC
* Renamed XHC1->XHC

That fixed USB 3.0 devices previously were not working (only when plugging them on USB 2.0 socket)

Must create a custom SSDT for USBInjectAll.kext to stay within the 15-port limit. Then remove the port limit patch.
EHCx renames are nonsense for your hardware (your chipset has no ECHI controller).

>> - VirtualBox tends to interfere with USB. remove.
>> - you have a bunch of things hooking your USB (Google Chrome, Plex Tuner Service)

That's crap but I guess I will have to live with it?

Remove VirtualBox. See laptop FAQ for Chrome. No idea about Plex. For you to research.
 
Yes.
You need to re-extract/re-patch ACPI.
Then avoid making any changes to BIOS options, BIOS version, or installed hardware that would change those addresses... otherwise, you'll need to do it again.

The alternative is to avoid putting patched files in ACPI/patched by patching with Clover config.plist options and additional SSDTs, which is known as "hotpatch."

Ok. I re-patch ACPI, add XCPM type PM X86 plugin looks loaded. Don't touch anything on BIOS, I am still getting slow boot.
Here is my new files;
 

Attachments

  • Archive.zip
    2.9 MB · Views: 108
Ok. I re-patch ACPI, add XCPM type PM X86 plugin looks loaded. Don't touch anything on BIOS, I am still getting slow boot.
Here is my new files;

Problems:
- HECI should be renamed IMEI
- FixRegions_10000000 is a bad idea (remove).
- you should NOT be dropping CPU SSDTs 'CpuPm' and 'Cpu0Ist' (remove from DropTables)
- probably you should remove config.plist/CPU/QPI setting
- remove 15-port limit patch from config.plist
- you might want to trim not needed efi drivers from drivers64UEFI
- kexts not installed correctly... all kexts you need must be installed to the system volume

You should analyze your kernel logs from start of boot to get an idea as to what might be taking a while.
 
hi guys, hi @RehabMan
i am concerned about the slow boot time with high sierra and my nvme 960 evo.
with sierra it was a lot faster.
i'm attaching here my files, hope someone could give me some tips.
 

Attachments

  • file alex.zip
    2.7 MB · Views: 111

Attachments

  • files.zip
    3.3 MB · Views: 94
  • bootlog dpci manager.zip
    9.4 KB · Views: 86
sorry. i hope now there is everything you need. thank your for your help.
ps my boot time is about 55 seconds (from clover screen to login screen)

Problems:
- no files in ACPI/origin (you forgot to press F4).
- port limit patch is in use (not a good idea, use it only to discover the ports, then create custom SSDT for USBInjectAll to stay within the 15-port limit). You're injecting 17 ports, which is two ports over the limit.
- EC0->EC patch is a bad idea if you have ECDT (and native EC is not named EC). No idea because you forgot to attach native ACPI/origin
- kexts are not installed correctly (all kexts you need should be installed to the system volume, typically /L/E). config.plist/SystemParameters/InjectKexts should be Detect.

You should also check your kernel logs to see what might be taking so long...
 
Hey, im having a slow startup too :-/
I dont use a NVMe, but an SSD.
Upgrade to HS went fine, and all seems to be working as it did before the update, except for the slow reboot.

Im sorry if this is not the place to address this, but i figured since the other guys also have the same issue.

I have attached the files.

Thanks
 

Attachments

  • Archive.zip
    3.3 MB · Views: 101
Last edited:
Hey, im having a slow startup too :-/
I dont use a NVMe, but an SSD.
Upgrade to HS went fine, and all seems to be working as it did before the update, except for the slow reboot.

Im sorry if this is not the place to address this, but i figured since the other guys also have the same issue.

I have attached the files.

Thanks

Problems:
- CPU PM not implemented (due to incorrect SMBIOS, see below)
- iMac14,2 is a poor choice for Sandy Bridge hardware (it is XCPM/X86PlatformPlugin, not ACPI_SMC_PlatformPlugin)
- Use Sandy MacMini instead
- lots of 3rd party system software installed
- old Ethernet kext
- might try an HFS+J install instead of APFS
 
Status
Not open for further replies.
Back
Top