Intel GPUs on Omarchy

Intel graphics is the quietest GPU class on Omarchy. On a machine where an Intel iGPU is the only GPU and drives the panel, 4.0.4 installs the driver stack itself and the only Intel-only gap cited on this page is a detection miss on Wildcat Lake. Most of the Intel reports cited here are really one of two other things: a hybrid Intel plus NVIDIA laptop where session variables point at the wrong GPU, or a Panther Lake panel fighting its kernel flags.

Status on 4.0.4

Intel-only desktops and laptops: works. Mesa handles rendering, intel-media-driver handles VA-API decode and encode, and vulkan-intel handles Vulkan. The installer scripts below and the 3.6.0 release notes, which added intel-media-driver by default and describe Intel laptops driving Thunderbolt displays, are the enablement. Of the 293 issues matched to this component, the ones cited on this page are hybrid machines, Panther Lake panels, or the Wildcat Lake detector miss in issue #11958.

Hybrid Intel plus NVIDIA: partial, and worse on 4.0.1 and later than it was on 4.0.0. See the hybrid section below and /hardware/hybrid-gpu/.

Panther Lake (Core Ultra X series, Arc B-series iGPU): mostly works, with active regressions. The 3.6.0 and 3.7.0 release notes carry the enablement work, from the custom kernel patches to fred=on. This is the newest silicon Omarchy ships for, and the kernel it ships changed twice in a month. 4.0.3 moved the Panther Lake kernel to 7.2.3, and 4.0.4 made the bespoke linux-omarchy kernel the default boot entry for everyone.

Old Intel (GMA era, pre-2014): unknown. The installer still matches it and installs the legacy libva-intel-driver rather than the modern stack, but no report in the cited set covers a GMA machine either way.

What Omarchy does automatically

The installer runs these on every machine during ISO finalization, through omarchy-apply-system and omarchy-apply-hardware. omarchy-update does not rerun that phase; existing installs get the pieces that matter through migrations, such as the 3.8.3 Vulkan backfill.

• install/hardware/intel/video-acceleration.sh looks at the lspci display line and installs intel-media-driver, libvpl and vpl-gpu-rt when the name matches HD Graphics, UHD Graphics, Xe, Iris, Arc or Panther Lake. GMA-era parts get libva-intel-driver instead.
• install/hardware/vulkan.sh installs vulkan-intel when a display-class Intel device is present. This is what made Zed start on Intel machines after issue #1441.
• install/hardware/intel/fred.sh writes /etc/limine-entry-tool.d/intel-panther-lake-fred.conf with fred=on when omarchy-hw-intel-ptl matches. Added in 3.7.0.
• install/hardware/intel/thermald.sh and intel/lpmd.sh install and enable thermald and intel-lpmd on Intel laptops, which shapes how hard the iGPU is allowed to run.
• install/hardware/asus/fix-asus-ptl-display-backlight.sh adds xe.enable_dpcd_backlight=1, and asus/fix-asus-ptl-b9406-display.sh adds xe.enable_panel_replay=0. Both are gated to the ASUS ExpertBook B9406 and Zenbook UX5406AA, and the scripts say other models still need confirmation.
• install/hardware/intel/ipu7-camera.sh installs intel-ipu7-camera for MIPI webcams. That stack has its own problems, covered at /hardware/webcam/.

The detectors behind this are omarchy-hw-intel (CPU vendor), omarchy-hw-intel-ptl (a Panther Lake display device), omarchy-hw-vulkan and omarchy-hw-hybrid-gpu. There is also omarchy-hw-dell-xps-oled, which matches an LG OLED panel on a Panther Lake XPS, but nothing in the 4.0.4 tree calls it.

One thing Omarchy does that hurts Intel: default/hypr/nvidia.lua sets LIBVA_DRIVER_NAME=nvidia, NVD_BACKEND=direct and __GLX_VENDOR_LIBRARY_NAME=nvidia whenever a GSP-era NVIDIA GPU is present, and autostart.lua exports that into the systemd user environment for every app.

Known problems

The hybrid VA-API regression is the big one. On 4.0.0 the o.shell_succeeds helper always returned false inside Hyprland, so nvidia.lua never fired and hybrid machines got the right driver by luck. 4.0.1 fixed the helper, the variables took effect, and browser video broke. SisyphusOfCorinth pinned that in issue #8215; reporters there confirm it on Raptor Lake, Meteor Lake, Arrow Lake and AMD iGPU pairings. emshiarla’s issue #8328 shows the detector only asks whether an NVIDIA chip exists, not whether it drives a display, and jtmorris measured roughly six extra watts in a browser on an XPS 14 9440 because the dGPU never runtime-suspends. Issue #8989 adds the battery cost on a Legion 7i Pro and notes that an override placed before the Omarchy require loses to it, which is the documented load order, so it has to go after. PR #7851 proposes gating the variable on a sysfs hybrid check; it is still open.

Panther Lake panels are the second cluster. Issue #5573 was a 3.x case where a stale xe.enable_panel_replay=0 line survived in /etc/default/limine and froze the XPS 14 internal panel after 3.7.0; spencerbull reproduced it and a hotfix went out as v3.7.1. Issue #12188 is the current one: after 4.0.4 makes linux-omarchy 7.2.5 the default, brightness keys on a Dell XPS 14 write successfully to intel_backlight but the panel does not change, and booting the stock linux 7.2.3 entry restores it. That symptom matches what the ASUS backlight script describes, so xe.enable_dpcd_backlight=1 is worth testing there, but nobody has confirmed it on an XPS.

Displays that stay dark after idle are the third. Issue #12147 traces it into omarchy-brightness-display: the script skips the DPMS enable when Hyprland already reports every monitor lit, and a failed atomic commit can leave that flag stuck true while the panel has no signal. Issue #12152 is the external-only variant on a Haswell i915 Vostro, recovered with a VT switch; the reporter traced it to an Aquamarine 0.15.0 regression that leaves a stale CRTC behind on disconnect, fixed upstream but not yet in the packaged Aquamarine.

Older i915 resume failures are still open in issue #5695, with PHY and transcoder timeouts on Meteor Lake after the 7.0 kernel bump.

Known issues

Issue	Models	Status	Fixed in
#8215 NVDEC routed to the dGPU, corrupt browser video	Hybrid Intel or AMD iGPU plus NVIDIA	Open, workaround known	Not yet, PR #7851 open
#8328 NVIDIA env forced when the dGPU drives no display	Comet Lake plus RTX 2070, Dell XPS 14 9440	Open	Not yet
#8989 NVIDIA env session-wide, dGPU cannot suspend	Lenovo Legion 7i Pro, Arrow Lake	Open	Not yet
#4901 Chromium acceleration needs manual flags	Raptor Lake Iris Xe plus RTX 3050	Open	Not yet
#11958 Wildcat Lake skipped by the detector regex	Wildcat Lake iGPU	Open	Not yet
#12188 Backlight dead on linux-omarchy 7.2.5	Dell XPS 14 DA14260, Arc B390	Open	Not yet
#12147 Monitor stays off after idle wake	Arrow Lake plus RTX 5070, external LG 4K	Open	Not yet
#12152 External-only HDMI unusable after wake	Dell Vostro 5470, Haswell i915	Open, Aquamarine fix upstream	Not yet
#11943 Hardware cursor stops rendering	Legion 7 16IRX9, i915 plus RTX 4060, 4 monitors	Open	Not yet
#5695 Black image and long freeze on resume	ThinkPad P1 Gen 7, Meteor Lake	Open	Not yet
#5573 Internal panel frozen after upgrade	Dell XPS 14, Panther Lake	Fixed	3.7.1
#1441 Zed fails to start, no Vulkan driver	Any Intel iGPU	Fixed	3.8.3 backfill migration; fresh installs via vulkan.sh
#6985 4.0 ISO install aborts before vulkan.sh runs	Any machine with a Synaptics-named input device	Fixed	4.0.1 ISO, PR #7236

Fixes that work

Work in this order.

1. Find out what you actually have. Run lspci -k | grep -A3 -E 'VGA|3D|Display' and journalctl -k | grep -E 'i915|xe '. Then run vainfo and systemctl --user show-environment | grep -E 'LIBVA|GLX|NVD'.
2. If LIBVA_DRIVER_NAME is nvidia and your panel hangs off the Intel iGPU, that is your bug. In ~/.config/hypr/hyprland.lua, after require("default.hypr.omarchy"), add hl.env("LIBVA_DRIVER_NAME", "iHD") and hl.env("__GLX_VENDOR_LIBRARY_NAME", "mesa"). It must come after the require, or it is overwritten. To avoid a relogin, also run systemctl --user set-environment LIBVA_DRIVER_NAME=iHD and restart the browser. Reporters in issue #8215 confirmed this on several chassis.
3. If vainfo finds no driver at all, install the packages by hand: sudo pacman -S intel-media-driver libvpl vpl-gpu-rt libva-utils. On Wildcat Lake this is expected until issue #11958 lands.
4. If a Vulkan app refuses to start, check pacman -Q vulkan-intel. Installing it is the whole fix.
5. On Panther Lake with a panel problem, read /etc/default/limine and /etc/limine-entry-tool.d/. Remove any xe.enable_psr=0 or xe.enable_panel_replay=0 line left over from before 3.6, run sudo limine-update, and reboot.
6. If the trouble started with 4.0.4, boot the older linux entry once from the Limine menu. The 4.0.4 migration keeps the previous kernel installed for exactly this. If the problem disappears, say so in a linux-omarchy issue.
7. If a display stays dark after idle, force a real transition: hyprctl dispatch 'hl.dsp.dpms({ action = "disable" })' then the same with enable.
8. For Chromium specifically, see /fix/chromium-flicker-hardware-acceleration/.

Report it

Run omarchy debug, or omarchy-debug directly if the group command is missing on your build. It writes /tmp/omarchy-debug.log with inxi -Farz, dmesg, the current boot journal at warning level and up, and your package list. Use omarchy debug --print to read it first, and --no-sudo if you would rather leave dmesg out.

Attach that, plus the exact output of lspci -k for the display device, vainfo, systemctl --user show-environment, uname -r, and which Limine entry you booted. For hybrid machines, say which connector your panel is on and whether the dGPU exposes any DRM connectors at all, because that is the distinction the current detector misses. Hardware reports for this site go to /hardware/submit/.

Related

/hardware/hybrid-gpu/, /hardware/nvidia/, /hardware/multi-monitor/, /hardware/suspend-sleep/, /hardware/dell-xps-14-2026/, /fix/chromium-flicker-hardware-acceleration/, /fix/cursor-invisible-or-wrong-size/, /releases/v4.0.4/, /reference/commands/omarchy-debug/. Manual chapter: Monitors.