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

[DimitrisG]

I tried the CPUFriend posted by @zanzy, as it matches my CPU. The CPU package power at idle, as reported by Intel Power Gadget, is the same, at about 1.5W, for me. This is measured with the display sleeping, which gives me the lowest reading, naturally.

I lost only a very slight amount in Geekbench score, but still, it appears that, yes, it's true there aren't power savings to be had from lower clocks, past a certain point. Perhaps it's not universally the case- I don't know. I only have 2 cores, and some of you have 4, which is a pretty big difference.

Anyway, this testing led me to the discovery that, for this machine at least, the "pmset -b displaysleep" setting isn't persisting. It resets to 10 minutes after every reboot. I'd like to find a fix for this! Though I guess I could just kluge it with a startup script.

My opinion is that even cpu running at lowest freq , battery consumption remains the same if we don't lower the voltage the cpu is working at this freq.

 

[Johann7]

Did you use pmset to disable hibernation? (hibernatemode = 0)

Yeah, I have disabled hibernate mode as per the original post using the pmset commands.
The output of pmset -g is:

standby              0
standbydelaylow      10800
halfdim              1
hibernatefile        /var/vm/sleepimage
proximitywake        0
gpuswitch            2
powernap             0
disksleep            0
standbydelayhigh     86400
sleep                0 # I've disabled this for stability, but would like to sue
autopoweroffdelay    28800
hibernatemode        0
autopoweroff         0
ttyskeepawake        1
displaysleep         0
tcpkeepalive         1
Standby Battery Threshold 50
lidwake              1

 

[Johann7]

Hi Mark,
Thanks for linking this. I'll give a try with the AppleALC and remove the Voodoo one and look into the sleep specific issues.
Thanks!

So, I did those fixes, and while initially, it only recognized the line-in for audio input, it eventually saw that I had a built-in mic, and it appears the input levels are fixed.

And for sleep, I reenabled it, and, perhaps it was the alcid=56 boot option, or the use of the AppleALC kext, but it seem to be working! This is a huge improvement!
Many thanks!

 

[kylec]

Again, I'm interested if anyone else finds the USB C to be not working with USB 3 devices? I've found this thread now: https://www.tonymacx86.com/threads/usb-c-hotplug-not-working.223534/post-1629770

For myself, setting RMV to one is all it takes for reliable USB-C hotplug

The change mentioned is included in @the-darkvoid 's SSDT-TYPC.dsl file, and my bdmesg output shows this SSDT being successfully loaded, yet in MaciASL I don't see any _RMV method on RP01.PXSX. Only the renamed XRMV from the Clover hot-patching. I also tried a simplified SSDT-TYPC with just that one method, returning one, but the result was the same.

Previously, I've always static patched, so I'm not sure what to do here.

Anyway, this testing led me to the discovery that, for this machine at least, the "pmset -b displaysleep" setting isn't persisting. It resets to 10 minutes after every reboot. I'd like to find a fix for this! Though I guess I could just kluge it with a startup script.

@RehabMan, if you have any idea on either one of these, thanks.

 

[jkbuha]

Hi all,

I've updated my configuration based on the i7-8550U chipset, with amended CPUFriend configuration as well.

Points to note:
1) The closest Apple hardware to this config is the MBP15,2 (i7-8559U with no GPU), so I've rebased the config based on this chipset.
2) Previously I was averse to using CPUFriend/HWP but it seems that Intel has improved a thing or two with the 8th gen. On optimal configuration I now get 5000+ single and 15000+ multi core results on GB4.

Feel free to use and amend as required.

 

[mashny]

@the-darkvoid I need your help with optimising the i7-8550U vectors.

I upgraded my motherboard into a i7-8550U w/ 16GB @ 2133Mhz. Surprisingly I can boot without CPUFriend, and it appears that I have most of the P-states as well, which means that XNU power management is working to some extent on 10.14.2.

I'm trying to optimise the CPU power profile now to maximise the performance under the 15W TDP. Not sure if I should use CPUFriend or not, or even Speed-Shift. Have you ever benchmarked the performance using Geekbench or any other tool? Any advice here would be appreciated.

Did you just transplant a motherboard from a 8550u system? If so, could you pass the part #?

 

[jkbuha]

Did you just transplant a motherboard from a 8550u system? If so, could you pass the part #?

Yes and it's relatively easy to do. The 93x0 motherboards are understandably small and low on external connections, so easy to replace.

The part I ordered with 16GB and an i7-8550U came with 2133MHz of LPDDR3. This is the part code:

MJ08X OEM DELL XPS 13 9360 Motherboard i7-8550U 3.8GHz CPU 16GB RAM LA-D051P

That should set you back around $300-450 excluding relevant taxes. Cheaper if you get lucky or hustled.

Apparently there also are some 2400MHz boards in the wild, but I don't know if they have a different part code.

 

[the-darkvoid]

@jkbuha, nice upgrade. I'm still waiting on my screen part, so currently not using the XPS 9360.
Hence no updates on my git. I expect it to arrive in the next few days.

Edit: Finally. My XPS 9360 is fixed and operational again. I did look at other laptops in the market before deciding to repair, but no major upgrades since this is already i7-8550u. Only becomes worth looking at a next laptop once 32GB hits 13"/14" inch factor and maybe Ice Lake or beyond CPUs.

 

[the-darkvoid]

Has anyone had a problem with the screen's hinge? It became loose last year and I made Dell fix it (well, the 'technician' replaced the whole screen instead of fixing the hinge) and now the problem has recurred, but now the laptop is out of warranty... sigh...

Tips? (slightly off-topic, but oh well)

is a great guide on how to replace your hinges only.

 

[egore]

VoltageShift

------------------------------------------------------

VoltageShift Info Tool

------------------------------------------------------

CPU voltage offset: -250mv

GPU voltage offset: -250mv

CPU Cache voltage offset: -75mv

System Agency offset: 0mv

Analogy I/O: 0mv

OC mailbox cmd failed

Digital I/O: 0mv

CPU BaseFreq: 2000, CPU MaxFreq(1/2/4): 4000/4000/3700 (mhz)

CPU Freq: 0.6ghz, Voltage: 0.5081v, Powerkg 1.99w /core 0.16w,Temp: 35 c

Above is the current read of my machine. CPU Cache cannot go lower without a freeze. CPU and GPU voltage certainly seem like they can go lower, but the script doesn't allow me to.

Is it officially determined that the BIOS needs to have overclocking enabled to make these adjustments actually take effect?

It does behaviorally seem like these adjustments are working, since lower CPU Cache values create a freeze of the machine.

Thanks much!