LeCoreOS 11 0.8.5 — latency setup and comparisons

cooked by LeSamsky

Zero input-to-display delay is not possible. This release adds reversible configuration controls, not a measured zero-delay result. The creator's current report of poor responsiveness has not been diagnosed on the PC. Registry matching cannot establish faster aim or lower end-to-end delay.

Start here

Close games and Control Center. Extract LeCoreOS-11-0.8.5-Update.zip and run Update.cmd from the original gaming administrator account. This app update preserves your current tuning.
Read Processor scheduling choices. Each explicit Apply is optional. The existing Apply competitive base button remains available separately; it is not run by this updater. Restart after any tuning Apply or Restore. Orange means configured values match; it is not measured latency.
Keep your gaming monitor directly connected to the gaming GPU and play from that display. Test with OBS projector/preview closed on the gaming PC. A capture/preview path is a separate delay source that Windows registry keys cannot eliminate. Keep your audio/capture routing tools if required, then compare the same scene with them running.
Select the monitor's maximum supported refresh rate in Windows Display settings. Use current device/chipset/controller drivers from the device vendors. Avoid changing drivers, overclocks and latency experiments in one test.

Competitive base

Reapplies the existing Windows Game Mode/capture preset: Game Mode On, background game recording Off.
Activates the existing LeCoreOS 11 Max Performance configuration. The v0.7.0 AC controls and one-hour AC/DC display/sleep timers are retained. Other power schemes are exported before the existing Max action removes them. The plan's label does not establish a latency advantage over every other plan.
Uses the already allowlisted optional DiagTrack and MapsBroker service controls. Controller, audio, network, anti-cheat, security and SysMain services remain protected.
Adds an explicit MMCSS comparison: SystemResponsiveness 10, Games Scheduling Category Medium, Games Priority 6. The original registry values and a .reg export are saved before mutation, with readback and rollback on a failed write. Restart before judging the comparison. It affects threads that use the relevant MMCSS task; benefit is not established for every game.

The MMCSS originals live in %ProgramData%\LeGameOS\LatencyControls.json; registry exports are in LatencyRegistryBackups. Gaming → Restore MMCSS restores values owned by this comparison. Later outside changes are retained with a warning. Recovery → machine recovery also invokes latency recovery. The existing user-settings and power recovery tools restore their respective originals. Keep all journals and power-plan exports.

Review of the requested tweak list

Proposal	Finding and implementation
SystemResponsiveness = 0	Microsoft documents values below 10 as clamped to 20. The comparison uses 10 instead of claiming zero reserves all CPU time for the game.
GPU Priority = 8; SFIO Priority = High	Microsoft lists both fields as unused. No new overrides are written for them.
Games category High plus priority 6	High is intended for Pro Audio and treats the numeric priority as 2. The optional MMCSS comparison uses Medium / 6.
Win32PrioritySeparation = 0x28	No universal current Windows 11 gaming advantage was established. The existing value is kept until an explicit scheduling Apply. The new selector offers twelve reversible comparisons; see PROCESSOR-SCHEDULING.html.
NetworkThrottlingIndex = ffffffff	This is not a guarantee of lower game ping or input delay. No new unverified Windows 11 networking override is added. Use the existing Ethernet comparisons and router SQM guide and measure loaded network latency separately.
Ultimate/A2 power scheme	The existing explicit Max controls already provide AC CPU minimum/maximum 100%, EPP 0, PCIe link saving Off and USB selective suspend Off where supported. No arbitrary third-party A2 plan is fabricated.
Disable all core parking / CPU idle	AC core unparking is an opt-in comparison below. CPU idle states, SMT, thread affinity and timer/BCD configuration are not altered.
Disable SysMain and Xbox services	Microsoft advises keeping SysMain; it supports caching rather than Windows Search indexing. Gaming/controller dependencies are retained. DiagTrack and MapsBroker are already optional targets.

Optional core-parking comparison

Gaming → Compare AC unparking sets only the existing active Max plan's CPMINCORES to 100%, using Microsoft's documented API/SDK identifier. It records the prior AC percentage first, validates the new value, and offers Restore core parking. It does not change battery values or processor idle states. The action is blocked while a supported game is open and on identified Ryzen 9 X3D models where chipset scheduling may rely on parking. Your Ryzen 7 9800X3D can use the comparison when the Max plan is active.

This may change heat, power and frame-time behavior. Keep it only if repeated tests improve your setup. Normal Windows core parking is not proof of a latency fault. Applying Max Performance again does not automatically enable this experiment.

Fortnite / VALORANT / NVIDIA

The existing game and driver actions are retained. Apply the corresponding profile, then confirm these choices in the game menu:

Fortnite: Performance rendering, graphics Low with view distance Far; V-Sync and motion blur Off; Reflex On + Boost on your RTX 5070. Save supported in-game choices in the app after closing the game.
VALORANT: Low/Off graphics, AA None / filtering 1x, V-Sync Off, Reflex On + Boost on NVIDIA. Close the game and Riot client before saving settings.
Keep frame generation/Smooth Motion off for an initial competitive-latency comparison. The app does not guess private game/driver enum values to control these features. Reflex is set in game; it takes precedence over NVIDIA driver low-latency mode.
Compare uncapped with Reflex against a stable in-game cap if frame-time spikes persist. Do not automatically stack driver, RTSS and in-game caps. A tear-free G-SYNC/V-Sync configuration is a different tradeoff from minimum-delay V-Sync Off.
For compatible DX10/11 borderless/windowed games, Windows 11 windowed optimizations use flip presentation to reduce frame latency. The app includes a link to the Windows settings; no blanket Fullscreen Optimization/MPO/DWM disablement is applied.

For your controller, compare a direct wired connection or the manufacturer's 2.4 GHz receiver against Bluetooth, using its supported polling settings and current firmware. No HID/USB overclock filter or GameInput service disablement is installed.

Verify on your PC

Use the same resolution, scene/replay route and game version. Warm up shaders first. Record three comparable runs before and after one change. Record average FPS, frame-time spikes and the game's supported render/PC-latency statistic separately. That statistic is not automatically controller-to-pixel delay; end-to-end measurement needs a suitable camera or latency measurement device. Test with streaming/projector routing both enabled and disabled. Revert a comparison that is slower, hotter or unstable.

Gaming → Save latency configuration report records local configuration, not an FPS/latency measurement. The native Windows UI preflight validates controls and themes, not game responsiveness. No physical Windows PC or games were available in the build environment.

Primary references — checked October 16, 2026

Microsoft MMCSS
Microsoft CPMinCores
Microsoft Win32 SDK GUID declarations
Microsoft service guidance — IoT guidance is used to identify service roles, not to claim desktop gaming benchmarks.
Microsoft windowed optimizations
NVIDIA latency guide — published 2020; current Windows presentation guidance is checked separately above.
Epic Performance Mode
Epic controller input troubleshooting