[Guide] Dell XPS 13 9360 on MacOS Sierra 10.12.x - LTS (Long-Term Support) Guide

[bozma88]

How can you tell if HWP is enabled?

AppleIntelInfo reports HWP as enabled but pmset -g everything |grep -i 'cycle' does not report any darkwake activity...

Could it be that darkwake=0 works as a kernel argument?

Then it is enabled, and maybe darkwakes are not happening. Did you tune the frequencyvectors? There's no vanilla match for 7500U.

 

[jkbuha]

Yes I tuned the frequency vectors myself for i7-7500U for both the MacBook9,1 and MacBookPro13,2 profiles.

I've attached my modded kext for reference.

 

[fv316]

Yes I tuned the frequency vectors myself for i7-7500U for both the MacBook9,1 and MacBookPro13,2 profiles.

I've attached my modded kext for reference.

The MacBook9,1 profile works pretty well, what changes did you make to the plist? I'm especially interested in what tuning you did to the frequency vectors and if you discovered which bit is responsible for maximum frequency?

 

[bozma88]

The MacBook9,1 profile works pretty well, what changes did you make to the plist? I'm especially interested in what tuning you did to the frequency vectors and if you discovered which bit is responsible for maximum frequency?

I think that is mandatory to tweak the vectors for the sole reason that our CPU has a boost frequency of 3500Mhz and there's not a direct match in the X86PlatformPlugin. Look here: http://www.insanelymac.com/forum/to...hift-enable-with-full-power-management/page-2

 

[jkbuha]

That is correct. In fact to get to 3500MHz I used a combination of tables, and built on the work of a number of others more capable than myself, such as @syscl.

 

[jkbuha]

Some further optimisation and testing using MB9,1+hotpatch+HWP. I've attached the results.

I have run this extensively for 48h and not experienced any darkwakes with this X86 profile but would appreciate if someone else verifies it as well.

 

[bozma88]

I created a SSDT to inject deep-idle, use-msi and nvme-LPSR-during-S3-S4 properties in RP09. I don't know whether they provide power savings, but now the properties are there in IOreg.
And I noticed that pci-aspm-default is enabled, too (was even before deep-idle injection, although), and it's active in L0s and L1 states (0x2 value). That's good!

I think I'll be able to tell if it makes a difference when I'll do a thermography of the NVMe drive, next week.

Other thing: A good reason to switch to Macbook13,x SMBIOS is that MacBook9,1 does not have thunderbolt in its USB-c connector. That may prevent thunderbolt devices to run correctly. Since I don't have any TB3 device, I cannot test, and I'm not sore it would be hot-swappable, as well, as that is currently the issue with USB3.1 (C) port.

 

[bozma88]

Some further optimisation and testing using MB9,1+hotpatch+HWP. I've attached the results.

I have run this extensively for 48h and not experienced any darkwakes with this X86 profile but would appreciate if someone else verifies it as well.

Please post your full config, so I'll test too.
Also, have a read here: https://www.tonymacx86.com/threads/...nvme-disk-corruption-during-darkwakes.223847/

EDIT: for some reason, the linked thread is waiting approval by moderators.

 

[jkbuha]

Sure - here you go. Let me know if you see the same results.

I'd like to try your NVME power settings but seeing as I'm still on 512b I'll refrain for now, in case deep-idle triggers some form of dark wake.

 

[bozma88]

Sure - here you go. Let me know if you see the same results.

I'd like to try your NVME power settings but seeing as I'm still on 512b I'll refrain for now, in case deep-idle triggers some form of dark wake.

FYI, Macbook9,1 frequencyvectors plist is corrupted in your archive