SteamOS battery optimization with measured energy per frame
Scope
Battery optimization is workload reduction while preserving an explicit experience target. Publish a measured range and method, never a timeless “hours of battery” promise.
Do not compare runs with different starting charge, charging state, battery temperature, background downloads, or scene duration. Those differences can exceed the tuning effect.
Baseline
Start at a recorded charge band on battery power, disable downloads, keep brightness and wireless state fixed, use one repeatable scene, and log device model, battery health if available, game/OS build, settings, refresh/cap, power profile, ambient temperature, duration, start/end charge, and average power.
Change one variable
Test one step at a time: FPS cap, matching refresh mode, GPU-heavy settings/resolution, TDP/power limit, brightness, HDR, wireless/background activity, then optional per-device controls. Leave shader compilation and first-run setup outside the comparison window.
Measurements
| Metric | Record | Reason |
|---|---|---|
Energy | Start/end charge or energy plus duration | Direct consumption evidence |
Power | Average and observed range | Faster comparison than full discharge |
Experience | Average FPS, 1% low, frame-time graph | Prevents buying runtime with stutter |
Environment | Brightness, radios, temperature, fan/noise | Makes runs comparable |
Repeat equal-duration runs and estimate runtime only from a stable, representative window. Label estimates and never extrapolate a menu or idle scene to full gameplay.
Side effects
Lower caps can increase latency; low TDP can create frame-time spikes; lower brightness harms outdoor readability; disabling radios breaks online play/cloud sync; HDR changes both image and power; a full shader rebuild can temporarily consume more power and stutter.
Rollback
Restore the original per-game profile, cap/refresh, TDP/automatic controls, brightness, HDR, and radio state. Restart the game and confirm the baseline scene returns to its original frame-time and power range.
Battery-health boundary
Runtime tuning does not repair battery wear. For long-term device storage, follow Valve's current battery storage-mode procedure for the exact Deck model. Do not improvise charge-controller or calibration procedures from unrelated hardware.
Frequently asked questions
Can one battery-life number apply to every Steam Deck game?Can one battery-life number apply to every Steam Deck game?
Does lowering refresh rate always save more power?Does lowering refresh rate always save more power?
Sources
Version history
- 2026-07-15: Phase 5 measured battery-optimization workflow published.