What this guide covers

Machine-specific fixes for the ASUS ProArt P16 (2025) (AMD Ryzen AI CPU + NVIDIA RTX dGPU) running Fedora Linux.

For everything that is not specific to this machine — NVIDIA driver install, Cardwire GPU switching, and asusctl power management, see:

• GPU Guide
• ASUS Laptop Control (asusctl)

Overnight Battery Drain while asleep

The symptom: close the lid on a full battery, wake up to a dead, warm laptop. On our unit this drained ~7% per hour asleep — a full battery gone in one night.

The touchpad (ASCF1A01:00) fires a constant interrupts (about 15/s). During modern standby (s2idle), the interrupts woke the machine from its deepest sleep state. Measured over a 6-hour sleep, the hardware only truly slept 53% of the time.

The fix: tell the system the touchpad is not allowed to wake the machine. The lid and the power button still wake it; tapping the touchpad no longer does (which most people never relied on anyway).

• Silence the touchpad as a wakeup source

  Create one udev rule so the setting applies on every boot:

  sudo tee /etc/udev/rules.d/90-touchpad-no-wakeup.rules <<'EOF'
  ACTION=="add", SUBSYSTEM=="i2c", KERNEL=="i2c-ASCF1A01:00", ATTR{power/wakeup}="disabled"
  EOF

  And apply it immediately without rebooting (this command only disables touchpad wakeup for one boot):

  echo disabled | sudo tee /sys/bus/i2c/devices/i2c-ASCF1A01:00/power/wakeup

• Verify

  Sleep the laptop for at least half an hour, wake it, and check how long the hardware really slept:

  grep . /sys/power/suspend_stats/last_hw_sleep
  # output is in microseconds — divide by 1,000,000 for seconds

  It should now be 95% or more of the time the lid was closed.

Results on our unit (AMD AI HX370, NVIDIA 4060, 32 GB)

Before any fixes: ~6.8% battery per hour asleep, warm chassis, 53% hardware-sleep residency.

After this fix: under 1% per hour and over 99.8% residency — a full night of sleep now costs under 10% battery.

How the culprit was found (technical detail)

Credit: Debugged with Claude

• journalctl showed one continuous suspend the whole night, so the machine was asleep — it just was not sleeping deeply.
• /sys/power/suspend_stats/last_hw_sleep showed hardware-sleep residency far below the suspend duration (53%).
• Enumerating /sys/class/wakeup/wakeup*/ by event_count pointed at i2c-ASCF1A01:00 with 582,000+ events within 8 hours.
• The storm itself (15 interrupts/s even while awake) looks like an i2c-hid quirk on this board; a future BIOS/firmware update may calm it. Disarming it as a wakeup source is enough to fix sleep either way.

Contributors

aier9500

Changelog

7/26/26, 4:46 PM

View All Changelog

• 0a483-Improved ProArt P16 guide with simpler language.on 7/26/26
• a6e96-Update to simplify wording, also to credit Claude.on 7/26/26
• 5f34b-Polish after removing redundant Nvidia fix.on 7/26/26
• 146a1-Sleep drain ProArt P16 and hybrid solutionson 7/26/26