The addition of the firmware in 20250311 seems to allow the NPU to shutdown correctly... I observed package power on my SoC drop by a few hundred mWs (with idle power approaching that of Windows at ~0.6w vs ~1w prior to the firmware update), despite cat /sys/class/pci_bus/0000\:00/device/0000\:00\:0b.0/power_state reporting D3hot in both instances
Kernel logs prior
saturn kernel: intel_vpu 0000:00:0b.0: enabling device (0000 -> 0002)
saturn kernel: intel_vpu 0000:00:0b.0: [drm] *ERROR* ivpu_fw_request(): Failed to request firmware: -2
saturn kernel: intel_vpu 0000:00:0b.0: [drm] ivpu_hw_power_down(): NPU not idle during power down
saturn kernel: intel_vpu 0000:00:0b.0: probe with driver intel_vpu failed with error -2
Kernel logs with updated firmware
saturn kernel: intel_vpu 0000:00:0b.0: enabling device (0000 -> 0002)
saturn kernel: intel_vpu 0000:00:0b.0: [drm] Firmware: intel/vpu/vpu_40xx_v0.0.bin, version: Jan 15 2025*NPU40xx*ci_tag_ud202504_vpu_rc_20250115_1905*ae83b65d01ccb4696594af0cafffea50a52520da
saturn kernel: intel_vpu 0000:00:0b.0: [drm] Scheduler mode: OS
saturn kernel: [drm] Initialized intel_vpu 1.0.0 for 0000:00:0b.0 on minor 0
The addition of the firmware in 20250311 seems to allow the NPU to shutdown correctly... I observed package power on my SoC drop by a few hundred mWs (with idle power approaching that of Windows at ~0.6w vs ~1w prior to the firmware update), despite
cat /sys/class/pci_bus/0000\:00/device/0000\:00\:0b.0/power_statereporting D3hot in both instancesKernel logs prior
Kernel logs with updated firmware