Dell XPS 15 9560 (Early 2017)

TheOneJester · 2017-05-11 20:55:34

Finally nvidia-381.22-1 is out in Extra repo. Anybody get a chance to try this yet?

EDIT: I just tried this without changing any kernel parameters and it appears for me that I can finally run optirun glxgears without any error messages.

Update: After rebooting a couple times and trying to use optirun, I've started to get: [ERROR]Cannot access secondary GPU - error: Could not load GPU driver

Update 2: I found out that some aggressive power saving may have caused that previous error. I must have power plugged in and reboot before I start the bumblebeed.service. The previous error will go away. My suspicion is that acpi_rev_override=1 must be set for this behavior to work correctly.

Last edited by TheOneJester (2017-05-11 23:59:19)

hedgepigdaniel · 2017-05-11 23:27:58

simo8989 wrote:

Nha.. that's not the problem, at least from htop seems all right. The thing is that I am trying on Fedora right now with kernel 4.10.14 and then I compile the kernel with the correct flag on yes but the situation does not change when then I boot with acpi_rev_override=1. The laptop almost melts before and after that and also the GPU is OFF, in the first case with original kernel sometimes and with custom kernel always but everything is hot and the fans spin constantly, I had those problems as peaks in temperature (80 C) actually for some minutes and not for all the time. I hope it's a Fedora problem and I will soon try with arch, btw which DE are you using? Gnome on Wayland?
I had Ubuntu 17.04 before this test and that kernel has the flag on y by default. With that Distro I had good results in power cosumption but never less than 14.5 W from powertop analysis. I am starting to think that the problem it's my computer.. it's very wired beacause on windows runs like a charm, with almost 9 h of battery life and the fans never spinning, apart when I charge the laptop.

80 is about the temperature I would expect if the CPU was at almost 100% load for a short time. Sounds to me like something is using the CPU (maybe a kernel bug doing it which prevents it from showing up in userspace tools?). Idk but if the temperature is 80 and the GPU is off the the CPU is having a party.

Maybe try running powertop while its happenning and have a look at all the CPU stats (idle, C state usage, processes etc). For me when idle the cores are 95% in C10SKL and the package in C8 (pc8), no process uses any significant CPU, CPU frequency is ~1100Mhz, and there are 25-50 wakeups per second.

Last edited by hedgepigdaniel (2017-05-11 23:32:22)

hedgepigdaniel · 2017-05-14 01:28:50

TheOneJester wrote:

Finally nvidia-381.22-1 is out in Extra repo. Anybody get a chance to try this yet?
...
Update 2: I found out that some aggressive power saving may have caused that previous error. I must have power plugged in and reboot before I start the bumblebeed.service. The previous error will go away. My suspicion is that acpi_rev_override=1 must be set for this behavior to work correctly.

I gave it a go and it worked but I found that gaming performance was significantly worse than with 375.66

Maybe the EFI turns on the GPU when rebooting with the power in, and otherwise leaves it in whatever state the OS left it? I think bbswitch is independent of the nvidia driver so I don't think a new nvidia driver would remove the need for acpi_rev_override

Ausprit · 2017-05-15 17:10:13

Any news on how the main kernel could help XPS15 to work? While building my own kernel sounds fun, it is high maintenance if I want to keep up with new versions. Any idea if the laptop will ever be supported without having to build a custom kernel?
As I understand it, they refused a patch, but I was wondering if another one was in the work or if custom kernels is the way to go.

zacariaz · 2017-05-15 17:49:36

Ausprit wrote:

Any news on how the main kernel could help XPS15 to work? While building my own kernel sounds fun, it is high maintenance if I want to keep up with new versions. Any idea if the laptop will ever be supported without having to build a custom kernel?
As I understand it, they refused a patch, but I was wondering if another one was in the work or if custom kernels is the way to go.

Yes, and how about the other available kernels? I hear that lts does something right, but then there is also the zen kernel, not to speak of those available through the AUR, ck kernels and so on and so forth.

Last edited by zacariaz (2017-05-15 18:03:00)

simo8989 · 2017-05-17 21:12:04

hedgepigdaniel wrote:

simo8989 wrote:
Nha.. that's not the problem, at least from htop seems all right. The thing is that I am trying on Fedora right now with kernel 4.10.14 and then I compile the kernel with the correct flag on yes but the situation does not change when then I boot with acpi_rev_override=1. The laptop almost melts before and after that and also the GPU is OFF, in the first case with original kernel sometimes and with custom kernel always but everything is hot and the fans spin constantly, I had those problems as peaks in temperature (80 C) actually for some minutes and not for all the time. I hope it's a Fedora problem and I will soon try with arch, btw which DE are you using? Gnome on Wayland?
I had Ubuntu 17.04 before this test and that kernel has the flag on y by default. With that Distro I had good results in power cosumption but never less than 14.5 W from powertop analysis. I am starting to think that the problem it's my computer.. it's very wired beacause on windows runs like a charm, with almost 9 h of battery life and the fans never spinning, apart when I charge the laptop.
80 is about the temperature I would expect if the CPU was at almost 100% load for a short time. Sounds to me like something is using the CPU (maybe a kernel bug doing it which prevents it from showing up in userspace tools?). Idk but if the temperature is 80 and the GPU is off the the CPU is having a party.
Maybe try running powertop while its happenning and have a look at all the CPU stats (idle, C state usage, processes etc). For me when idle the cores are 95% in C10SKL and the package in C8 (pc8), no process uses any significant CPU, CPU frequency is ~1100Mhz, and there are 25-50 wakeups per second.

I found the time to install arch Linux finally and I literally followed your guide in the wiki apart for 1 step. In arch linux with the compiled custom kernel the situation is much better but still I continue to get those 15w of power consumption in powertop and then I thought about the only different thing that I do from you... that is to always install the Linux os into a usb key because I never changed the operating mode of the hard drive from raid to ahci, this because I basically wanted to test arch very well before installing it alongside windows 10. Do you think that this can be the factor that causes my strange power consumption of 15w on idle ? Thank you

simo8989 · 2017-05-19 19:34:25

@hedgepigdaniel Hi! Look, this is my powertop that also says that the discharge rate is around 15w, as you can see everything is how you describe (including the idle in C10-SKL state) except for the Package states! Here I get a really high C3 and I think this is what cause my problem I get a battery life around 3 and half hours and the temperatures starts really good at 39C but then they rise until 50 C .... the fans are turning on and off on battery during normal firefox browsing. I don't understand what it could be, also because there is no significant process that uses the CPU. Thank you for your help.

          Package   |             Core    |            CPU 0       CPU 4
                    |                     | C0 active   0.8%        0.1%
                    |                     | POLL        0.0%    0.9 ms  0.0%    0.0 ms
                    |                     | C1E-SKL     0.1%    0.7 ms  0.0%    1.0 ms
C2 (pc2)   24.6%    |                     |
C3 (pc3)   66.6%    | C3 (cc3)    0.0%    | C3-SKL      0.0%    0.0 ms  0.0%    0.2 ms
C6 (pc6)    0.0%    | C6 (cc6)    0.2%    | C6-SKL      0.1%    0.9 ms  0.2%    4.6 ms
C7 (pc7)    0.0%    | C7 (cc7)   97.1%    | C7s-SKL     0.0%    1.5 ms  0.0%    0.0 ms
C8 (pc8)    0.0%    |                     | C8-SKL      4.9%    3.3 ms  1.9%    7.4 ms
C9 (pc9)    0.0%    |                     | C9-SKL      0.0%    0.0 ms  0.0%    0.0 ms
C10 (pc10)  0.0%    |                     | C10-SKL    92.9%   34.3 ms 97.7%   70.4 ms

                    |             Core    |            CPU 1       CPU 5
                    |                     | C0 active   0.5%        0.1%
                    |                     | POLL        0.0%    0.0 ms  0.0%    0.0 ms
                    |                     | C1E-SKL     0.0%    0.4 ms  0.0%    0.0 ms
                    |                     |
                    | C3 (cc3)    0.0%    | C3-SKL      0.0%    0.3 ms  0.0%    0.0 ms
                    | C6 (cc6)    0.7%    | C6-SKL      0.8%   10.1 ms  0.0%    0.3 ms
                    | C7 (cc7)   97.9%    | C7s-SKL     0.3%    3.4 ms  0.0%    0.0 ms
                    |                     | C8-SKL      1.6%    3.9 ms  0.8%    4.8 ms
                    |                     | C9-SKL      0.4%    4.5 ms  0.0%    0.0 ms
                    |                     | C10-SKL    96.3%   63.1 ms 98.9%  101.3 ms

                    |             Core    |            CPU 2       CPU 6
                    |                     | C0 active   0.1%        0.3%
                    |                     | POLL        0.0%    0.0 ms  0.0%    0.0 ms
                    |                     | C1E-SKL     0.0%    0.0 ms  0.0%    0.1 ms
                    |                     |
                    | C3 (cc3)    0.0%    | C3-SKL      0.0%    0.0 ms  0.0%    0.3 ms
                    | C6 (cc6)    0.1%    | C6-SKL      0.0%    0.0 ms  0.1%    0.8 ms
                    | C7 (cc7)   98.7%    | C7s-SKL     0.0%    0.0 ms  0.0%    0.0 ms
                    |                     | C8-SKL      1.0%    4.8 ms  4.2%    5.7 ms
                    |                     | C9-SKL      0.3%    5.5 ms  0.0%    0.0 ms
                    |                     | C10-SKL    98.4%   91.1 ms 95.2%   29.1 ms

                    |             Core    |            CPU 3       CPU 7
                    |                     | C0 active   0.3%        1.1%
                    |                     | POLL        0.0%    0.0 ms  0.0%    0.0 ms
                    |                     | C1E-SKL     0.0%    0.4 ms  0.1%    0.5 ms
                    |                     |
                    | C3 (cc3)    0.0%    | C3-SKL      0.0%    0.0 ms  0.0%    0.0 ms
                    | C6 (cc6)    0.1%    | C6-SKL      0.1%    1.1 ms  0.1%    0.8 ms
                    | C7 (cc7)   96.1%    | C7s-SKL     0.0%    1.6 ms  0.0%    0.0 ms
                    |                     | C8-SKL      4.8%   11.3 ms  4.2%    5.2 ms
                    |                     | C9-SKL      0.0%    0.9 ms  0.0%    0.0 ms
                    |                     | C10-SKL    94.0%   61.6 ms 93.3%   27.7 ms

                    |             GPU     |
                    |                     |
                    | Powered On  3.3%    |

As you can see there are many wake up per second here, and I have no Idea what it could be the cause

The battery reports a discharge rate of 17.3 W
The power consumed was 308 J
The estimated remaining time is 4 hours, 48 minutes

Summary: 340.1 wakeups/second,  12.3 GPU ops/seconds, 0.0 VFS ops/sec and 11.6% CPU use

                Usage       Events/s    Category       Description
             56.1 ms/s      84.2        Process        [PID 7822] /usr/bin/gnome-shell
              1.5 ms/s      42.5        Process        [PID 9239] /usr/lib/firefox/firefox -contentproc -greomni /usr/lib/firefox/omni.ja -appomni /usr/lib/firefox/browser/omni.ja -a
             59.5 µs/s      25.9        kWork          intel_atomic_helper_free_state
              3.7 ms/s      23.9        Process        [PID 9163] /usr/lib/firefox/firefox
            449.6 µs/s      25.2        Process        [PID 309] [i915/signal:0]
             22.4 ms/s      13.3        Process        [PID 9217] /usr/lib/firefox/firefox -contentproc -greomni /usr/lib/firefox/omni.ja -appomni /usr/lib/firefox/browser/omni.ja -a
            406.2 µs/s      20.9        Interrupt      [6] tasklet(softirq)
              0.7 ms/s      20.4        Timer          tick_sched_timer
             18.1 ms/s      11.9        Process        [PID 8216] /usr/lib/gnome-terminal/gnome-terminal-server
             81.5 µs/s      18.9        Timer          intel_uncore_fw_release_timer
              2.5 ms/s       9.2        Interrupt      [127] xhci_hcd
             61.0 µs/s       7.8        kWork          intel_unpin_work_fn
              0.0 µs/s       7.8        kWork          intel_mmio_flip_work_func
             83.1 µs/s       7.2        Process        [PID 7] [rcu_preempt]
             51.1 µs/s       5.8        kWork          gen6_pm_rps_work
            660.9 µs/s       4.2        Interrupt      [164] i915
              2.2 ms/s       1.2        Interrupt      [7] sched(softirq)
            115.2 µs/s       1.8        Process        [PID 9169] /usr/lib/firefox/firefox
             24.1 µs/s       1.8        Process        [PID 154] [usb-storage]
            171.4 µs/s       1.3        Interrupt      [3] net_rx(softirq)
            170.9 µs/s       1.0        Process        [PID 536] /usr/bin/NetworkManager --no-daemon
             33.2 µs/s       1.0        Process        [PID 9224] /usr/lib/firefox/firefox -contentproc -greomni /usr/lib/firefox/omni.ja -appomni /usr/lib/firefox/browser/omni.ja -a
             69.5 µs/s       1.0        Process        [PID 7980] /usr/lib/gnome-settings-daemon/gsd-color
             58.3 µs/s       1.0        Process        [PID 7530] /usr/lib/gnome-settings-daemon/gsd-color
             12.1 µs/s       0.9        kWork          pci_pme_list_scan
              8.6 µs/s       0.8        kWork          ieee80211_iface_work
              4.0 µs/s       0.8        Timer          sched_rt_period_timer
            530.5 µs/s       0.6        Process        [PID 704] /usr/bin/gnome-shell
             93.9 µs/s       0.7        Process        [PID 7915] /usr/lib/telepathy/mission-control-5
             91.3 µs/s       0.7        Process        [PID 7977] /usr/lib/gnome-settings-daemon/gsd-sharing
              1.6 ms/s      0.10        Interrupt      [0] HI_SOFTIRQ
              9.5 µs/s       0.7        kWork          vmstat_shepherd
             65.0 µs/s       0.5        kWork          delayed_fput
             21.8 µs/s       0.5        kWork          i915_gem_idle_work_handler
             15.7 µs/s       0.5        Process        [PID 8413] gpg-agent --homedir /home/simone/.gnupg --use-standard-socket --daemon
             18.8 µs/s       0.4        kWork          __i915_gem_free_work
              0.0 µs/s       0.4        kWork          disk_events_workfn
             44.3 µs/s       0.4        Process        [PID 7518] /usr/lib/geoclue2/geoclue -t 5
              0.7 ms/s      0.10        Process        [PID 63] [khugepaged]
              9.1 µs/s       0.3        Process        [PID 9556] powertop
              6.6 µs/s       0.3        kWork          i915_gem_retire_work_handler
             77.3 µs/s      0.30        Process        [PID 8585] /usr/lib/dleyna-renderer/dleyna-renderer-service
            683.1 µs/s      0.00        Timer          hrtimer_wakeup
             30.0 µs/s      0.25        Process        [PID 7846] /usr/bin/Xwayland :0 -rootless -noreset -listen 4 -listen 5 -displayfd 6
              5.4 µs/s      0.25        Process        [PID 15] [watchdog/1]

Last edited by simo8989 (2017-05-19 23:48:13)

zacariaz · 2017-05-19 19:42:12

Could you try the lts kernel? Someone reported it making great improvements to power usage.

osho741 · 2017-05-19 23:21:14

I reported decent battery life with LTS and all the way up to 4.10.10-NVMe give me decent powertop figures with TLP and no acpi override set.(meaning kernel isn't customized in any way) and the 1050 card is still on. All I can figure is my Desktop Environment which is a openbox/XFCE being so resource conservative to begin with it is probably helping battery life

Last edited by osho741 (2017-05-19 23:23:10)

hedgepigdaniel · 2017-05-20 01:35:26

simo8989 wrote:

@hedgepigdaniel Hi! Look, this is my powertop that also says that the discharge rate is around 15w, as you can see everything is how you describe (including the idle in C10-SKL state) except for the Package states! Here I get a really high C3 and I think this is what cause my problem I get a battery life around 3 and half hours and the temperatures starts really good at 39C but then they rise until 50 C .... the fans are turning on and off on battery during normal firefox browsing. I don't understand what it could be, also because there is no significant process that uses the CPU. Thank you for your help.

As you can see there are many wake up per second here, and I have no Idea what it could be the cause

Hmm yes that does look different. My CPU package states looked like this:

C2 (pc2)	1.1%
C3 (pc3)	3.1%
C6 (pc6)	0.0%
C7 (pc7)	0.0%
C8 (pc8)	88.8%
C9 (pc9)	0.0%
C10 (pc10)	0.0%

Also at the bottom is sais "Powered on: 0.3%"

Only guessing here but maybe if you're frequently accessing a mount point on a usb drive there is some optimisation that doesn't happen as it would with NVME? Or maybe because the USB is slow I/O takes a long time and the CPU stays on for longer?

The wakeups also look very different - mine sais 37 wakeups per second. For both of us it seems the wakeups for all the processes don't quite add up to the total wakeups, so I'm guessing they are only ones caused by userspace processes. Looks kind of similar in terms of the ratios between everything - just that all of yours are waking up much more lol

simo8989 · 2017-05-20 08:27:38

@hedgepigdaniel Thank you! we can clearly see the difference now in your C8 state for package. I also was wondering what GPU 3.3% (in my case) means? I guess is related to the intel GPU (hopefully).
As soon as I have time I will install Arch on the nvme drive and hope to get the values as you. In the meanwhile even more interesting could be if you try to install a random distro on a usb stick and see if you get my 15W power discharge also applying all your suggestions in the wiki page you created. If you want to try it I suggest Ubuntu since it has already a kernel compiled with the ACPI flag that we need. Thank you!
Simone

Last edited by simo8989 (2017-05-20 08:28:34)

hedgepigdaniel · 2017-05-21 03:04:32

simo8989 wrote:

@hedgepigdaniel Thank you! we can clearly see the difference now in your C8 state for package. I also was wondering what GPU 3.3% (in my case) means? I guess is related to the intel GPU (hopefully).
As soon as I have time I will install Arch on the nvme drive and hope to get the values as you. In the meanwhile even more interesting could be if you try to install a random distro on a usb stick and see if you get my 15W power discharge also applying all your suggestions in the wiki page you created. If you want to try it I suggest Ubuntu since it has already a kernel compiled with the ACPI flag that we need. Thank you!
Simone

My GPU was 0.3% powered on - I think that's the integrated GPU. dmesg would tell you if the discrete one was turned on.

Haha, I think I'll stick with my SSD

tj8892 · 2017-05-21 11:42:17

Hi all
I posted earlier about how following someone's guide to install nvidia-beta-dkms and a few other packages. Since nvidia driver 380 made it into the repo, I've been using that just fine instead of the beta. I get, with gnome, wifi connected (I'm far away from the router), brightness at 50%, and firefox+ some terminals open about 10W of use, leading to 5-9hours of battery depending on what I do. (1080p screen). The post recommended a slew of kernel parameters that I added to systemd-boot (I'm using pcie_port_pm=off acpi_backlight=vendor acpi_osi=Linux acpi_osi=! acpi_osi="Windows 2009" rcutree.rcu_idle_gp_delay=1). I was curious as to what these actually do, because despite having them, dmesg practically vomits ACPI errors at me. I found that bumblebee powering off the nvidia card did not work at all, causing power to go up to 20W, and optirun didn't work unless I had pcie_port_pm=off, but I'm seeing no differences with the others on or off. "ACPI Warning: \_SB.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20160930/nsarguments-95)" is spat out about twenty times in dmesg, but I don't have any issues when using just pcie_port_pm=off. What do acpi_backlight=vendor, acpi_osi=Linux, acpi_osi=!, acpi_osi="Windows 2009", and rcutree.rcu_idle_gp_delay=1 actually do, and should I be using them? Also, what powersaving stuff are you using? I've got just tlp enabled right now. All stock options in config, except I've replaced CPU_HWP_ON_BAT=balance_power with balance_performance, because the default option, balance_power caused major lag on gnome when going to the overview (replacing that option definitely makes gnome snappier, and doesn't seem to drain the battery more at all.)

gilbertw1 · 2017-05-23 18:22:30

I just updated to 4.11.2 and my discharge rate jumped from 12-14W to 7-9W!! The estimated battery life is averaging about 11-12 hours, whereas it was around 7-8 hours before.

I didn't make any changes other than updating linux.... is anyone else experiencing similarly awesome improvements?

j9ac9k · 2017-05-23 18:34:21

I just updated before, still seeing pretty high power usage, but more annoying, I can't adjust my monitor brightness :-/

simo8989 · 2017-05-23 21:18:37

gilbertw1 wrote:

I just updated to 4.11.2 and my discharge rate jumped from 12-14W to 7-9W!! The estimated battery life is averaging about 11-12 hours, whereas it was around 7-8 hours before.
I didn't make any changes other than updating linux.... is anyone else experiencing similarly awesome improvements?

Good news! do you compile a kernel 4.11.2 with the ACPI_REV_OVERRIDE_POSSIBLE? do you still use acpi_rev_override=1 as boot parameter ?

Last edited by simo8989 (2017-05-23 21:18:48)

simo8989 · 2017-05-23 22:32:47

Here is my power consumption on Idle with the kernel 4.11.2 (I get this only with the ACPI_REV_OVERRIDE_POSSIBLE). The thing I do not understand is why do I get those 100% stuck parameters for many of the Pci devices?

The battery reports a discharge rate of 9.03 W
The power consumed was 0.00 J

              Usage     Device name
             12.1%        CPU core
             12.1%        DRAM
             12.1%        CPU misc
              3.9 ops/s   GPU core
              3.9 ops/s   GPU misc
             25.1%        Display backlight
              0.0%        Radio device: btusb
            100.0%        USB device: xHCI Host Controller
            100.0%        USB device: Extreme (SanDisk)
            100.0%        USB device: xHCI Host Controller
            100.0%        USB device: USB Receiver (Logitech)
            100.0%        I2C Device (i2c-DLL07BE:01): DLL07BE:01
            100.0%        Radio device: ath10k_pci
            100.0%        PCI Device: Intel Corporation Sunrise Point-H PCI Express Root Port #15
            100.0%        PCI Device: Intel Corporation Skylake Processor Thermal Subsystem
            100.0%        PCI Device: Qualcomm Atheros QCA6174 802.11ac Wireless Network Adapter
            100.0%        PCI Device: Realtek Semiconductor Co., Ltd. RTS525A PCI Express Card Reader
            100.0%        PCI Device: Intel Corporation Sunrise Point-H USB 3.0 xHCI Controller
            100.0%        PCI Device: Samsung Electronics Co Ltd NVMe SSD Controller SM961/PM961
            100.0%        PCI Device: Intel Corporation Device 591b
            100.0%        PCI Device: Intel Corporation Sunrise Point-H PCI Express Root Port #1
            100.0%        PCI Device: Intel Corporation Sunrise Point-H Thermal subsystem
            100.0%        PCI Device: Intel Corporation Sunrise Point-H PCI Express Root Port #9
            100.0%        PCI Device: Intel Corporation Sunrise Point-H SATA Controller [AHCI mode]
            100.0%        PCI Device: Intel Corporation Sunrise Point-H PCI Express Root Port #2
              9.7 pkts/s  Network interface: wlp2s0 (ath10k_pci)
              0.0%        USB device: Integrated_Webcam_HD (CN045G28724876CKB0FTA01)
              0.0%        USB device: usb-device-0cf3-e300
              0.0%        Audio codec hwC0D2: Intel
              0.0%        Audio codec hwC0D0: Realtek
              0.0%        PCI Device: Intel Corporation Device a171
              0.0%        PCI Device: Intel Corporation Device 5910
              0.0%        PCI Device: Intel Corporation Sunrise Point-H PCI Express Root Port #13
              0.0%        runtime-INT3400:00
              0.0%        PCI Device: Intel Corporation Sunrise Point-H SMBus
              0.0%        PCI Device: Intel Corporation Sunrise Point-H PMC
              0.0%        PCI Device: Intel Corporation Sunrise Point-H Serial IO I2C Controller #1
              0.0%        PCI Device: Intel Corporation Sunrise Point-H LPC Controller
              0.0%        PCI Device: Intel Corporation Skylake PCIe Controller (x16)
              0.0%        PCI Device: Intel Corporation Sunrise Point-H CSME HECI #1
              0.0%        PCI Device: NVIDIA Corporation GP107M [GeForce GTX 1050 Mobile]
              0.0%        runtime-PNP0103:00
              0.0%        runtime-INT3403:00
              0.0%        runtime-INT33A1:00
              0.0%        runtime-microcode
              0.0%        runtime-MSFT0101:00
              0.0%        runtime-INT0800:00

Last edited by simo8989 (2017-05-23 22:45:14)

hedgepigdaniel · 2017-05-23 23:08:46

tj8892 wrote:

Hi all
What do acpi_backlight=vendor, acpi_osi=Linux, acpi_osi=!, acpi_osi="Windows 2009", and rcutree.rcu_idle_gp_delay=1 actually do, and should I be using them?

Not sure, but I think the bold text in the OP sais they are not necessary now that we know acpi_rev_override fixes the GPU problem. I don't use them myself.

hedgepigdaniel · 2017-05-23 23:11:21

simo8989 wrote:

Here is my power consumption on Idle with the kernel 4.11.2 (I get this only with the ACPI_REV_OVERRIDE_POSSIBLE). The thing I do not understand is why do I get those 100% stuck parameters for many of the Pci devices?

Same for me, very confusing. Especially for the SSD which is 100% on even though I'm pretty sure ASPT is working.

simo8989 · 2017-05-23 23:51:18

hedgepigdaniel wrote:

simo8989 wrote:
Here is my power consumption on Idle with the kernel 4.11.2 (I get this only with the ACPI_REV_OVERRIDE_POSSIBLE). The thing I do not understand is why do I get those 100% stuck parameters for many of the Pci devices?
Same for me, very confusing. Especially for the SSD which is 100% on even though I'm pretty sure ASPT is working.

Also I am pretty sure ASPT is working I checked it through the nvme-cli and unlike you I have the Samsung ssd. Also unlike you I do not need to specify the boot parameter for the nvme core because the sum of my biggest enlat+xlat is 8200 that does not exceed the default 25000.
Anyway in may case I get a decent power consumption now from powertop than on total idle is about 9-10 W and also the temperatures are low and they stay around 38-39 C when I browse Firefox.

Last edited by simo8989 (2017-05-23 23:52:58)

gilbertw1 · 2017-05-23 23:52:05

simo8989 wrote:

gilbertw1 wrote:
I just updated to 4.11.2 and my discharge rate jumped from 12-14W to 7-9W!! The estimated battery life is averaging about 11-12 hours, whereas it was around 7-8 hours before.
I didn't make any changes other than updating linux.... is anyone else experiencing similarly awesome improvements?
Good news! do you compile a kernel 4.11.2 with the ACPI_REV_OVERRIDE_POSSIBLE? do you still use acpi_rev_override=1 as boot parameter ?

I didn't compile the kernel, I'm just using the stock one. Here are the boot parameters I'm using:

GRUB_CMDLINE_LINUX_DEFAULT='pcie_port_pm=off acpi_backlight=vendor acpi_osi=Linux acpi_osi=! acpi_osi="Windows 2009" rcutree.rcu_idle_gp_delay=1'

simo8989 · 2017-05-23 23:54:37

gilbertw1 wrote:

simo8989 wrote:
gilbertw1 wrote:
I just updated to 4.11.2 and my discharge rate jumped from 12-14W to 7-9W!! The estimated battery life is averaging about 11-12 hours, whereas it was around 7-8 hours before.
I didn't make any changes other than updating linux.... is anyone else experiencing similarly awesome improvements?
Good news! do you compile a kernel 4.11.2 with the ACPI_REV_OVERRIDE_POSSIBLE? do you still use acpi_rev_override=1 as boot parameter ?
I didn't compile the kernel, I'm just using the stock one. Here are the boot parameters I'm using:
GRUB_CMDLINE_LINUX_DEFAULT='pcie_port_pm=off acpi_backlight=vendor acpi_osi=Linux acpi_osi=! acpi_osi="Windows 2009" rcutree.rcu_idle_gp_delay=1'

Thank you, as soon as I can I try out them and report back.

osho741 · 2017-05-24 01:28:00

I'm getting the same kind of numbers with stock kernel and TLP-git installed. I tried this kernel with acpi rev overide set in a custom build last night and could t get it to boot with bbswitch running. I may have been doing something wrong though

tj8892 · 2017-05-24 04:13:38

4.11 seems to work fine. Previously I was using linux-nvme, with the samsung pm961 SSD. Now I've switched from nvidia-dkms and bbswitch-dkms to just nvidia and bbswitch. My battery life has gone from good to straight up ridiculous since getting 4.11 on stock kernel. I get 6-7W with firefox open and basic web browsing on wifi, at half brightness leading to 11~ hours of battery. What other tweaks are you guys using? I think everyone should use the repository tlp package instead of tlp-git because the AUR package seems abandoned and it outdated, while the repo package is up to date. TLP enabled with default settings gets my battery down to 5W, but has a few annoying problems.

osho741 · 2017-05-24 04:54:22

I have no idea where I went wrong on it last night I got as far as bumblebee service going. I rebooted installed bbswitch and after the reboot I got stuck at boot. Are you just setting acpi-rev-overide on in the kernel config file, and installing bumblebee and bbswitch with a boot parameter set to pcie_port_pm=off ?

Last edited by osho741 (2017-05-24 04:59:27)

Arch Linux

#176 2017-05-11 20:55:34

Re: Dell XPS 15 9560 (Early 2017)

#177 2017-05-11 23:27:58

Re: Dell XPS 15 9560 (Early 2017)

#178 2017-05-14 01:28:50

Re: Dell XPS 15 9560 (Early 2017)

#179 2017-05-15 17:10:13

Re: Dell XPS 15 9560 (Early 2017)

#180 2017-05-15 17:49:36

Re: Dell XPS 15 9560 (Early 2017)

#181 2017-05-17 21:12:04

Re: Dell XPS 15 9560 (Early 2017)

#182 2017-05-19 19:34:25

Re: Dell XPS 15 9560 (Early 2017)

#183 2017-05-19 19:42:12

Re: Dell XPS 15 9560 (Early 2017)

#184 2017-05-19 23:21:14

Re: Dell XPS 15 9560 (Early 2017)

#185 2017-05-20 01:35:26

Re: Dell XPS 15 9560 (Early 2017)

#186 2017-05-20 08:27:38

Re: Dell XPS 15 9560 (Early 2017)

#187 2017-05-21 03:04:32

Re: Dell XPS 15 9560 (Early 2017)

#188 2017-05-21 11:42:17

Re: Dell XPS 15 9560 (Early 2017)

#189 2017-05-23 18:22:30

Re: Dell XPS 15 9560 (Early 2017)

#190 2017-05-23 18:34:21

Re: Dell XPS 15 9560 (Early 2017)

#191 2017-05-23 21:18:37

Re: Dell XPS 15 9560 (Early 2017)

#192 2017-05-23 22:32:47

Re: Dell XPS 15 9560 (Early 2017)

#193 2017-05-23 23:08:46

Re: Dell XPS 15 9560 (Early 2017)

#194 2017-05-23 23:11:21

Re: Dell XPS 15 9560 (Early 2017)

#195 2017-05-23 23:51:18

Re: Dell XPS 15 9560 (Early 2017)

#196 2017-05-23 23:52:05

Re: Dell XPS 15 9560 (Early 2017)

#197 2017-05-23 23:54:37

Re: Dell XPS 15 9560 (Early 2017)

#198 2017-05-24 01:28:00

Re: Dell XPS 15 9560 (Early 2017)

#199 2017-05-24 04:13:38

Re: Dell XPS 15 9560 (Early 2017)

#200 2017-05-24 04:54:22

Re: Dell XPS 15 9560 (Early 2017)

Board footer