Dell XPS 15 9570

From ArchWiki
Jump to navigation Jump to search
Note: This page far from completed. Some not mentioned items could be same as XPS 15 9560. Most of it also applies to the Precision 5530
Device/Functionality Status
Suspend Working
Hibernate Working
Integrated Graphics Working
Discrete Nvidia Graphics Modify
Wifi Working
Bluetooth Working
rfkill Working
Audio Working
Touchpad Working
Touchscreen Working
Webcam Working
Card Reader Working
Function/Multimedia Keys Working
Power Management Working
EFI firmware updates Working
Fingerprint reader[broken link: invalid section] Not working


Before installing it is necessary to modify some UEFI Settings. They can be accessed by pressing the F2 key repeatedly when booting.

  • Change the SATA Mode from the default "RAID" to "AHCI". This will allow Linux to detect the NVME SSD. If dual booting with an existing Windows installation, Windows will not boot after the change but this can be fixed without a reinstallation.
  • Change Fastboot to "Thorough" in "POST Behaviour". This prevents intermittent boot failures.
  • Disable secure boot to allow Linux to boot. Alternatively, you can choose to set up Secure Boot so that it works with Linux and you can take advantage of its potential security benefits.

Installation of Arch Linux can proceed normally. Refer to the Installation guide for more information.

Power Management


By default, the very inefficient s2idle suspend variant is incorrectly selected. This is probably due to the BIOS. The much more efficient deep variant should be selected instead:

 $ cat /sys/power/mem_sleep 
 [s2idle] deep
 $ echo deep|sudo tee /sys/power/mem_sleep
 $ cat /sys/power/mem_sleep 
 s2idle [deep]

To make the change permanent add mem_sleep_default=deep to your kernel parameters.

An easy way would be to add mem_sleep_default=deep to the GRUB_CMDLINE_LINUX_DEFAULT entry in /etc/default/grub:


Read more regarding the sleep variants on the kernel documentation [1].

Warning: Some users have reported a problem where the CPUs get stuck in a high power state after resuming from S3 (deep) suspension [2].


Works out of the Box see Power management/Suspend and hibernate


powertop is very efficient to manage power consumption. Run powertop --auto-tune and compare the Watt consumption variation (battery must be unplugged). powertop --auto-tune can be run at every boot.


kernel modules

The nouveau module is known to cause kernel panics and freezes on Dell XPS 15 9570.

One way to mitigate this would be by adding nomodeset to the kernel options. However it's better to completely disable it with the blacklist method (recommended):

blacklist nouveau
blacklist rivafb
blacklist nvidiafb
blacklist rivatv
blacklist nv

NVIDIA Optimus

The optimus configuration is a technology that allows an Intel integrated GPU and discrete NVIDIA GPU to be built into and accessed by a laptop. As the discret NVIDIA GPU card eats lots of power, we want to use the intergrated Intel card most of the time and activate/desactivate the NVIDIA GPU card only when required by a defined application. Due to a bug [3], the XPS 15 9570's secondary NVIDIA GPU cannot be powered down by bbswitch. However, standard Linux power management is able to properly turn off the card when the driver is unloaded.

Manually loading/unloading NVIDIA module

One solution [4] to the problem described above is to manually unload the nvidia module when not in use, which can be done as follows:


  • bumblebee - The main package providing the daemon and client programs.
  • nvidia - Official NVIDIA drivers for Linux.
  • powertop - Tool to verify power consumption.
  • unigine-valleyAUR - Graphic intensive application for testing

Make sure that bbswitch is uninstalled or at least disabled

Warning: To make optirun working flawlessly after enabling the NVIDIA GPU card, you have to verify that your regular user is in bumblebee group and bumblebeed.service is enabled and started. See Bumblebee#Installation.

bumblebee configuration, in the [driver-nvidia] section

# [driver-nvidia] section

Server X configuration for not auto adding gpu

Section "ServerFlags"
	Option "AutoAddGPU" "off"

ipmi modules are loaded together with nvidia and block its unloading. Create the two following files to disable them:

  install ipmi_msghandler /usr/bin/false
  install ipmi_devintf /usr/bin/false
install nvidia /bin/false

Now, add nvidia and ipmi in the modprobe.d blacklist to disable this functionality:

blacklist nouveau
blacklist rivafb
blacklist nvidiafb
blacklist rivatv
blacklist nv
blacklist nvidia
blacklist nvidia-drm
blacklist nvidia-modeset
blacklist nvidia-uvm
blacklist ipmi_msghandler
blacklist ipmi_devintf

Create 2 GPU management scripts for enabling and disabling discret NVIDIA graphical card [5]:
# allow to load nvidia module
mv /etc/modprobe.d/disable-nvidia.conf /etc/modprobe.d/disable-nvidia.conf.disable

# Remove NVIDIA card (currently in power/control = auto)
echo -n 1 > /sys/bus/pci/devices/0000\:01\:00.0/remove
sleep 1
# change PCIe power control
echo -n on > /sys/bus/pci/devices/0000\:00\:01.0/power/control
sleep 1
# rescan for NVIDIA card (defaults to power/control = on)
echo -n 1 > /sys/bus/pci/rescan
# someone said that modprobe nvidia is needed also to load nvidia, to check
# modprobe nvidia

modprobe -r nvidia_drm
modprobe -r nvidia_uvm
modprobe -r nvidia_modeset
modprobe -r nvidia

# Change NVIDIA card power control
echo -n auto > /sys/bus/pci/devices/0000\:01\:00.0/power/control
sleep 1
# change PCIe power control
echo -n auto > /sys/bus/pci/devices/0000\:00\:01.0/power/control
sleep 1

# Lock system form loading nvidia module
mv /etc/modprobe.d/disable-nvidia.conf.disable /etc/modprobe.d/disable-nvidia.conf

Create service locking NVIDIA GPU on shutdown A service which locks GPU on shutdown / restart when it is not disabled by script is necessary. Otherwise, on next boot nvidia will be loaded together with ipmi modules (even if blacklisted with install command) and it won't be possible to unload them then.

Description=Disables Nvidia GPU on OS shutdown

ExecStop=/bin/bash -c "mv /etc/modprobe.d/disable-nvidia.conf.disable /etc/modprobe.d/disable-nvidia.conf || true"


Reload systemd daemons and enable the disable-nvidia-on-shutdown service:

 sudo systemctl daemon-reload
 sudo systemctl enable disable-nvidia-on-shutdown.service

Allow gpu to poweroff on boot

w /sys/bus/pci/devices/0000:01:00.0/power/control - - - - auto

Finally, check that everything is well configured:

  1. Reboot and verify that nvidia is not loaded by running:
    lsmod | grep nvidia
    Should not return anything
  2. Disconnect / unplug charger and verify the power consumption with powertop is around 5W on idle (Dell XPS 15 4570, powertop --auto-tune previously launched, FHD screen with no touchscreen)
  3. Enable GPU by using the script
  4. Check that the GPU is loaded by using:
  5. If good, launch unigine-valley with optirun:
    optirun unigine-valley
    unigine-valley needs a GPU to work. Should activate the fans quickly.
  6. Close all nvidia applications and disable gpu with the script
  7. Check again power consumption, it should have gone back to a similar value as before

Letting bumblebee automatically unload the kernel module

Another possible way of shutting off the secondary GPU is to use a patched version of bumblebee that automatically unloads the nvidia kernel module when the card is not in use [6]. This patch has already been merged into the development branch of bumblebee [7] which can be obtained via bumblebee-gitAUR. It is also provided by bumblebee-forceunloadAUR, which is a modified version of bumblebee with this patch applied.


Make sure that AlwaysUnloadKernelDriver is set to true in the driver-nvidia section of your bumblebee configuration:

# [driver-nvidia] section

While the above option unloads the nvidia kernel module when not in use, Xorg will still automatically discover the secondary graphics card and attempt to enable it. In order to prevent this, the option AutoAddGPU should be disabled. However, this would lead to Xorg unable to detect the default integrated graphics card. The intel graphics card needs to be manually added to Xorg. This can be done by creating the following two configuration files in /etc/X11/xorg.conf.d:

Section "ServerFlags"
	Option "AutoAddGPU" "off"
Section "Device"
	Identifier	"Intel Graphics"
	Driver		"intel"
	# Uncomment line below if you are also having issues with backlight
	# Option		"Backlight"	"intel_backlight"

Set GPU power control to auto on boot so it will be powered off when the module is unloaded:

w /sys/bus/pci/devices/0000:01:00.0/power/control - - - - auto

Enable bumblebeed.service to start on boot:

 sudo systemctl enable bumblebeed.service

After restarting your system, you should be able to verify that everything is working properly:

  1. Verify that nvidia is not loaded on boot by running:
    lsmod | grep nvidia
    Should not return anything
  2. Check that secondary graphics card is functioning properly with optirun:
    optirun glxgears
  3. Don't close this window just yet, we need to check that the nvidia module is loaded correctly:
    lsmod | grep nvidia
    This time, it should be able to find nvidia in the loaded modules list
  4. After closing glxgears, check again to make sure that nvidia is properly unloaded:
    lsmod | grep nvidia
    Should not return anything once again
  5. At this point, the graphics card should be working correctly. You can further check the power consumption by disconnecting the charger and using Powertop, there should be a significant power difference when optirun glxgears is running and when it is not.



To get xbacklight working and not conflicting with NVIDIA optimus:

Section "Device"
	Identifier  "Intel Graphics"
        Driver      "intel"
        Option      "Backlight"  "intel_backlight"

NVRM: Failed to enable MSI; falling back to PCIe virtual-wire interrupts

Sometimes it happens after suspend/resume. GPU could work fine without MSI. [8]. You could disable MSI by adding the following in /etc/modprobe.d/nvidia.conf:

 options nvidia NVreg_EnableMSI=0

Built-in screen flickers or does not come on with Linux kernel 5.0.0 - 5.0.7

Some users reported that running Linux kernel 5.0.0 to 5.0.7 can cause the screen to flicker (or stay completely black) when booting up or running an X server, making the built-in display unusable (see *[9])

Currently, it seems that there are three possible workarounds :

Lock-ups when resuming from suspend with nvidia module

If your system locks up every time you resume from suspend with the following two lines in dmesg:

   [   42.447364] pci 0000:01:00.0: Refused to change power state, currently in D3
   [   46.896493] pci 0000:01:00.0: Refused to change power state, currently in D3

you need to do the following:

Into /etc/default/grub

   GRUB_CMDLINE_LINUX="nouveau.blacklist=0 acpi_osi=! acpi_osi=\"Windows 2015\" acpi_backlight=vendor mem_sleep_default=deep"

This solved the lock-ups for me. (internal nvidia bug number: 2589324, dell resolution)

Wifi and Bluetooth

These work well out of the box but you might need to update the firmware for better stability. For bluetooth, make sure you have everything installed as per the Bluetooth wiki page.

 $ lspci | grep Network
 3b:00.0 Network controller: Qualcomm Atheros QCA6174 802.11ac Wireless Network Adapter (rev 32)

(please add a line here if you detect something different)


ath10k module crashes after suspend

You may notice driver crashes after suspend/resume, which for the most part does not seem to impact the running system, with coredumps similar to:

kernel: WARNING: CPU: 6 PID: 27936 at drivers/net/wireless/ath/ath10k/mac.c:5746 ath10k_bss_info_changed+0xf96/0x1120 [ath10k_core]
kernel: Modules linked in: uhid algif_hash cmac rfcomm fuse ccm ipt_MASQUERADE nf_conntrack_netlink nfnetlink xfrm_user xfrm_algo iptable_nat nf_nat_ipv4>
kernel:  snd_hda_intel dcdbas x86_pkg_temp_thermal dell_smm_hwmon snd_hda_codec intel_powerclamp cfg80211 kvm_intel snd_hda_core input_leds snd_hwdep snd>
kernel:  vfio_mdev mdev vfio_iommu_type1 vfio kvm irqbypass intel_gtt i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops drm agpga>
kernel: CPU: 6 PID: 27936 Comm: kworker/u24:42 Tainted: G     U  W  OE     5.0.4-arch1-1-ARCH #1
kernel: Hardware name: Dell Inc. XPS 15 9570/0HWTMH, BIOS 1.8.1 02/01/2019
kernel: Workqueue: events_unbound async_run_entry_fn
kernel: RIP: 0010:ath10k_bss_info_changed+0xf96/0x1120 [ath10k_core]
kernel: Code: 24 8b 95 78 01 00 00 85 c0 0f 85 a7 00 00 00 89 d1 be 10 00 00 00 48 c7 c2 c0 b2 fa c0 4c 89 c7 e8 bf 68 00 00 e9 a5 f1 ff ff <0f> 0b 4c 89>
kernel: RSP: 0000:ffffb7a45422fcd0 EFLAGS: 00010282
kernel: RAX: 00000000fffffffe RBX: ffff98f6d44815a0 RCX: 0000000000000000
kernel: RDX: ffff98f6d4481938 RSI: ffffb7a45422fcf0 RDI: ffff98f6d81df418
kernel: RBP: ffff98f6d81df418 R08: 0000000000000001 R09: 0000000000000000
kernel: R10: 000000000000001f R11: 0000000000000000 R12: 0000000000000020
kernel: R13: ffff98f6d81df420 R14: ffff98f6d4482598 R15: ffff98f6d44815a0
kernel: FS:  0000000000000000(0000) GS:ffff98f6dc380000(0000) knlGS:0000000000000000
179 94%
kernel:  process_one_work+0x1eb/0x410
kernel:  worker_thread+0x2d/0x3d0
kernel:  ? process_one_work+0x410/0x410
kernel:  kthread+0x112/0x130
kernel:  ? kthread_park+0x80/0x80
kernel:  ret_from_fork+0x35/0x40
kernel: ---[ end trace 09ae3e174c178f98 ]---

Patched in some kernels and not others (which?), relevant links:

Bluetooth disappears (after suspend?)

Possibly related to the previous instability issue, sometimes the adapter seems to completely disappear. As reported here (thanks w.v.w., you can likely fix this by manually upgrading the firmware (to something newer than what's in linux-firmware:

1. Double check adapter (e.g. QCA6174, below)

 $ lspci | grep Network
 3b:00.0 Network controller: Qualcomm Atheros QCA6174 802.11ac Wireless Network Adapter (rev 32)

2. Confirm hardware and current firmware version (e.g. hw3.2, firmware RM.4.4.1.c3-00013-QCARMSWPZ-1, below)

 $ journalctl -b | grep ath10k | egrep 'firmware|qca'
 kernel: ath10k_pci 0000:3b:00.0: qca6174 hw3.2 target 0x05030000 chip_id 0x00340aff sub 1a56:1535
 kernel: ath10k_pci 0000:3b:00.0: firmware ver RM.4.4.1.c3-00013-QCARMSWPZ-1 api 6 features wowlan,ignore-otp,no-4addr-pad,raw-mode,mfp crc32 fc0096a8

3. Follow the instructions at

e.g. for the above adapter, download the latest firmware from and

 $ cd /lib/firmware/ath10k/QCA6174/hw3.0
 $ sudo cp firmware-6.bin firmware-6.bin.bak
 $ sudo cp ~/Downloads/firmware-6.bin_RM.4.4.1.c3-00013-QCARMSWPZ-1 firmware-6.bin

Either sudo rmmod ath10k_core ath10k_pci && sudo modprobe ath10k_pci (may not work, check dmesg) or reboot.

Touchpad and Touchscreen

Wacom touchscreen and Synaptics touchpad:

$ xinput
⎡ Virtual core pointer                          id=2    [master pointer  (3)]
⎜   ↳ Virtual core XTEST pointer                id=4    [slave  pointer  (2)]
⎜   ↳ SYNA2393:00 06CB:7A13 Touchpad            id=16   [slave  pointer  (2)]
⎜   ↳ Wacom HID 488F Finger                     id=15   [slave  pointer  (2)]
⎣ Virtual core keyboard                         id=3    [master keyboard (2)]

Both are i2c devices:

$ udevadm info /dev/input/mouse3 # touchscreen
P: /devices/pci0000:00/0000:00:15.0/i2c_designware.0/i2c-10/i2c-WCOM488F:00/0018:056A:488F.0006/input/input47/mouse3
N: input/mouse3
L: 0
E: DEVPATH=/devices/pci0000:00/0000:00:15.0/i2c_designware.0/i2c-10/i2c-WCOM488F:00/0018:056A:488F.0006/input/input47/mouse3
E: DEVNAME=/dev/input/mouse3
E: ID_PATH=pci-0000:00:15.0-platform-i2c_designware.0
E: ID_PATH_TAG=pci-0000_00_15_0-platform-i2c_designware_0
$ udevadm info /dev/input/mouse6 # touchpad
P: /devices/pci0000:00/0000:00:15.1/i2c_designware.1/i2c-11/i2c-SYNA2393:00/0018:06CB:7A13.0007/input/input38/mouse6
N: input/mouse6
L: 0
S: input/by-path/pci-0000:00:15.1-platform-i2c_designware.1-mouse
E: DEVPATH=/devices/pci0000:00/0000:00:15.1/i2c_designware.1/i2c-11/i2c-SYNA2393:00/0018:06CB:7A13.0007/input/input38/mouse6
E: DEVNAME=/dev/input/mouse6
E: ID_SERIAL=noserial
E: ID_PATH=pci-0000:00:15.1-platform-i2c_designware.1
E: ID_PATH_TAG=pci-0000_00_15_1-platform-i2c_designware_1
E: DEVLINKS=/dev/input/by-path/pci-0000:00:15.1-platform-i2c_designware.1-mouse

Thunderbolt docks


TB16 works fine if either Thunderbolt security is disabled in the BIOS or using bolt to temporarily authorize or permanently enroll Thunderbolt devices with Thunderbolt security activated. Various quirks are detailed on the Dell TB16 page.

EFI firmware updates

They are 2 main ways to update efi firmware:

  • through running linux session with Fwupd
  • through UEFI


 pacman -S fwupd

Then, look at:

 sudo fwupdmgr get-updates
 sudo fwupdmgr refresh
 sudo fwupdmgr update


Firmware images can be found at Dell support page as XPS_15_9570_X.Y.Z.exe files. In order to install:

  • Download the desired firmware from section "Dell XPS 15 9570 System BIOS"
  • Save it in /boot/EFI/Dell/Bios/ (this path may vary, depending on your installation)
  • Reboot the system, and enter the boot menu by pressing repeatedly F12 on Dell logo
  • Choose "Bios Flash Update"
  • Select the file previously saved, and start the process

The process will take about five minutes, during which the system will have some reboots and push fans at maximum speed. Finally the system will reboot normally.

Thermal management

Thermal design is poor in the 9570, primarily due to higher-end chips being used inside the original system design without compensating for extra heat. This impacts numerous areas:

  1. Performance: at higher temperatures, CPU cores are throttled down to avoid damage. At best your system will not run as fast as it can, and at worst (and quite commonly), slower than cheaper chips and with sluggish performance.
  2. Battery life is unnecessarily decreased.
  3. System longevity: running at constantly high temperatures will negatively impact equipment lifetime.
  4. User discomfort: uncomfortable heat and uncomfortable fan noise.

Fortunately these can all be improved significantly to get your system running powerfully and quietly.


You probably have a lot of dmesg output like this (for all CPUs), even with fairly light usage:

 kernel: CPU8: Package temperature above threshold, cpu clock throttled (total events = 8451)
 kernel: CPU8: Package temperature/speed normal

Use lm_sensors and do some stress testing (with stress and mprime) to see what's happening with CPU core temperatures and fanspeed under different loads.


See Undervolting CPU on the wiki. Reduces heat and extends battery life.

Possible configurations for /etc/intel-undervolt.conf:

  1. i9-8950HK* (last updated 2019-03-30)
 undervolt 0 'CPU' -140
 undervolt 1 'GPU' -75
 * Tested extensively for moderate use.  Was not stress tested for > 24 hours.  Anecdotal reports of up to -200.

This has a more minor impact than repasting but is significantly easier to do.

Repasting & padding

Significant improvements are possible by replacing the thermal grease used by Dell with better options available from Amazon, and adding some extra thermal padding. This might sound overwhelming but is well worth the effort, especially for newer CPUs. If you can't do this yourself consider finding a shop / technician who can do this for you. See the following article as the user comments below it for more info:


According to the author of the UltraBookReview article above:

Undervolting seems to reduce temps at max load by 7-10C, while repasting seems to reduce temps by between 4-10C depending on your original paste job and paste used.

From this forum post, much lower/softer fan speeds were needed to maintain the same temperatures, and temperature was a few degrees lower under similar loads. Fans were on less often and for less time when they were. Throttling was less prevalent and less severe.

Tips and Tricks

Systemd doesn't wait for Network

Few months ago systemd added "after= .. .." in /usr/lib/systemd/system/systemd-user-sessions.service

This causes systemd to wait for network connection at boot, you can modify this file to remove but it will be overwritten on systemd update. A better workaround is to add /etc/systemd/system/systemd-user-sessions.service with removed.

#  This file is part of systemd.
#  systemd is free software; you can redistribute it and/or modify it
#  under the terms of the GNU Lesser General Public License as published by
#  the Free Software Foundation; either version 2.1 of the License, or
#  (at your option) any later version.

Description=Permit User Sessions

ExecStart=/usr/lib/systemd/systemd-user-sessions start
ExecStop=/usr/lib/systemd/systemd-user-sessions stop