Skip to content

USB selective suspend and NVMe power states in Windows 11 25H2

On this page

Short answer

Section titled “Short answer”

USB selective suspend and NVMe idle power are configured by the power plan. The mechanism, the stornvme.sys handlers and the timeout ranges are confirmed. Changing them is justified to fix a specific device wake issue or to manage heat, but not to increase performance.

What was checked

Section titled “What was checked”

We checked where the USB and NVMe power settings are configured, which of them stornvme.sys reads, and which drive states they change.

Scope of research

Section titled “Scope of research”

Windows 11 25H2 build 26200.9168, USB hub and NVMe power paths. A specific power plan was checked; vendor plans and different drive models may differ.

Methodology

Section titled “Methodology”

Comparison of power plan settings with stornvme.sys handlers and verification of branches against device capabilities; for USB, cross-checking with public Microsoft documentation.

USB selective suspend

Section titled “USB selective suspend”

USB selective suspend is configured in the power plan, not by an ordinary global DWORD value in the registry. Windows can suspend an idle USB device and resume it on request.

Selective suspend reduces power consumption and laptop heat at idle. Disable it only to fix a reproducible disconnect or resume issue with a specific device. There is no basis to expect a permanent performance gain.

NVMe power settings

Section titled “NVMe power settings”

The settings are located in the storage subgroup of the power plan:

Setting Role Default in the researched plan
NVMe non-operational power state permission (noppme) allows/prohibits deep NVMe idle states enabled
primary idle timeout (idle1) timeout to primary idle state 200 ms
secondary idle timeout (idle2) timeout to secondary idle state 2000 ms
AHCI link power management (lpm) HIPM/DIPM link policy SATA/AHCI-specific

stornvme.sys registers handlers for power setting changes and applies noppme, idle1 and idle2 to a supported NVMe adapter. For both idle intervals, the documented power plan range is 0..60000 ms. Zero values prohibit the transition to the corresponding idle states, increasing drive power consumption at idle.

The lpm setting relates to AHCI/SATA link power management. In the NVMe service code of Windows 11 25H2, no separate handler for this setting was found.

Usefulness

Section titled “Usefulness”

On a desktop PC under constant load, an NVMe drive may rarely change power state. In that case, changing the timeout may not affect latency. On a laptop, disabling power management noticeably increases energy consumption and temperature. Verify the result on your own drive with the selected power plan.

What is confirmed

Section titled “What is confirmed”
  • The purpose of USB selective suspend and its binding to the power plan.
  • The NVMe settings subgroup and the 200/2000 ms values in the researched plan.
  • The handlers for noppme, idle1 and idle2 in stornvme.sys and the 0..60000 ms range.
  • The absence of a separate handler for lpm in the NVMe service code of Windows 11 25H2.

What is not confirmed

Section titled “What is not confirmed”
  • The effect on FPS, frame time or latency.
  • Behavior on other drive models, plans and USB controllers.
  • The benefit of disabling power management on a desktop PC.

Practical conclusion

Section titled “Practical conclusion”

Leave power management enabled. Disable selective suspend or increase the timeouts only to fix a reproducible issue with a specific device, and verify the result on your own drive.

Restoring the state

Section titled “Restoring the state”

Return the power plan values and reboot Windows if necessary. The new timeouts are applied the next time the device goes idle.

Limitations and sources

Section titled “Limitations and sources”

The conclusions apply to stornvme.sys Windows 11 25H2 and the checked power plan. It showed values of 200/2000; vendor plans may differ. The result also depends on the USB controller, firmware, NVMe model, AC or battery power, and PCIe power-saving settings.

For how to reproduce the dynamic part of the observations, see How to check it yourself.

The research and the tools used belong to the BoosterX developer, so the developer has a direct interest in the results. The methodology and the scope of applicability are described above, and the conclusions can be verified against open data and the listed public sources.

Public sources checked: 2026-09-02.

Change history

Section titled “Change history”
  • 2026-09-20: added a conflict-of-interest disclaimer and a link to self-verification of dynamic observations in the methodology.
  • 2026-09-02: first publication; confirmed the NVMe power handlers, timeout ranges and the boundaries of the effect on devices.