Scope: this documents the earlier 0.3.x base profile. LeCoreOS 11 0.8.0 adds Competition reductions and optional compact hosting; see COMPETITION.md. Statements about avoiding grouping describe the base profile alone. The latest requested idle band is 60–70, and your reported 0.3.2 result is 120. No new count or latency result is certified.

LeCoreOS 11 0.3.1 — lean profile and idle processes

Research checked 14 October 2026 UTC / 15 October in India. This is a configuration design and source review, not a Windows benchmark.

The target and what can be claimed

The earlier target was 50–60 idle processes, with a maximum of 70. The current 0.8.0 profile records your revised 60–70 band and checks it with the idle audit. It has not been achieved or verified on Windows here, and the package does not promise that the count will stay below 70. Earlier numerical estimates were planning guesses, not measurements of your PC.

The useful outcome is less unnecessary work while Fortnite and Valorant remain reliable. A process count is a partial view of that outcome. One busy process can disturb frame times more than many dormant ones. Closing a launcher with several helper processes can matter more than disabling ten services which were already stopped. That is an inference about what to measure, not an FPS claim for this package.

Microsoft documents service-trigger startup: services can wait without running until a device or another event requires them. Disabling a dormant service therefore need not remove an idle process. Service trigger events.

Microsoft also describes the historical change from shared to separate service hosts, with reliability and isolation benefits. The same service workload can therefore appear as very different process counts. That document concerns Windows 10 and supplies no current Windows 11 gaming benchmark. LeCoreOS 11 does not change service-host grouping or isolation to obtain a smaller counter. Service-host grouping.

What the new profile does

The scope is a private desktop used for your games and audio/capture setup. There are 19 exact system-service targets and five exact user-service templates. This is not a claim of 24 fewer processes. Targets absent from your Windows build are skipped; already-stopped or shared-host services may give no count reduction.

Selection is informed by Microsoft's specialized-device service catalogue. Its context is Windows IoT Enterprise, so it is not treated as blanket approval for every gaming desktop. The table above describes this playbook's choices; Windows game and driver testing is still required.

The new reductions check for a private desktop. They skip laptops, domain/cloud-joined systems and unavailable hardware/management information. Detected sensor devices keep their sensor services. Core devices, network discovery, Wi-Fi and Bluetooth are retained for your controller and two-PC setup. These checks limit scope; they cannot discover every application's undocumented dependency.

System-service changes are limited to recognized Win32 service types and expected startup values. Active or protected dependent services prevent a change. Original startup, delayed-start and running state are saved first. No service binaries are deleted and no security permission is weakened to overcome a refusal. ServiceController dependent-service API.

Mail, contact sync and recording

Per-user services have a template and session instances. The selected templates are configured with Set-Service; current instances are not forcibly killed. Restart or sign out to recreate them with the new startup configuration. The template changes are machine-wide and affect later user sessions. They can be restored through the backed-up machine settings; restart afterwards.

This disables Windows' own mail/calendar/contact services and Game DVR, while retaining credentials, notifications, device pairing and the separate Windows capture service. OBS, Voicemeeter and capture-card programs are kept. That separation follows the documented roles; compatibility on your actual hardware remains untested. Microsoft per-user services.

The app-removal feature also adds exact legacy Mail/Calendar, People, Maps, Cortana and Office Hub package names to the existing optional list. Packages absent from your account are ignored. Store, App Installer, Gaming Services, Xbox dependencies and system frameworks are retained. Removing an app that was not running is a feature reduction, not demonstrated CPU savings.

Known current-user Run entries for Steam, Epic Games Launcher, Riot Client, EA Desktop and Battle.net are disabled when their anchored names match. Start the launcher normally when you need it. Launchers and their installation/update services are not uninstalled. Vanguard startup is not targeted. Existing Discord/Brave startup reduction remains included. Original entries are journaled; scheduled tasks, machine-wide startup entries and different value names may still launch software. Microsoft Run keys.

Dependencies kept for gaming

The protected list covers OS startup and sign-in, RPC/COM, security, device installation, display, audio, controllers, network access, storage, servicing, licensing and game installation. Diagnostics and SysMain remain available. Only the explicit optional targets can be changed; arbitrary services are rejected.

For your 9800X3D/RTX 5070 setup, AMD/NVIDIA driver helpers, HID/USB, Bluetooth, GameInput, audio, OBS/capture and Voicemeeter are retained. Services marked optional for an unrelated device scenario can still be useful for this PC. Actual game launches and peripheral tests are the acceptance check.

• Valorant: Vanguard may require TPM 2.0, Secure Boot, HVCI/VBS and IOMMU, depending on the system. Those settings and anti-cheat services are preserved. Riot requirements.
• Fortnite: Epic requires Easy Anti-Cheat. EAC/BattlEye services are excluded from reductions. Epic EAC support.
• Warzone: TPM 2.0 and Secure Boot are required. RICOCHET and Windows security dependencies are retained. Activision requirements.
• FC 27: EA's applicable Secure Boot and Javelin requirements remain available. EA Secure Boot guidance.

No universal anti-cheat approval is claimed. Windows/game updates can add dependencies.

Measure the real count

1. Apply the selected profile and restart. Finish pending installations/updates.
2. Wait about three minutes after sign-in. Close foreground apps and launchers.
3. Open LeCoreOS 11 Control Center → 19. It takes 12 samples, approximately two seconds apart.
4. Read the minimum, median, maximum and average, plus whether every observed sample was at or below 70.
5. Repeat after another restart. Compare to a before-install baseline or recovered settings under the same conditions.

Reports are local under %LOCALAPPDATA%\LeGameOS\IdleAudit. They include process names/IDs, service names/states and the sampled counts. No process is excluded from the counter to improve the result; the audit's PowerShell process is included. Processes between samples can be missed, so even a passing audit establishes only the sampled window.

The CPU ranking estimates activity for readable processes present at both endpoints, normalized across logical processors. It excludes the idle pseudo-process and reused process IDs. It is not total Task Manager CPU usage. The service inventory is collected after sampling and may differ from the measured window.

If the maximum exceeds 70, the audit reports that result. It does not kill unknown services or block Windows process creation. Additional startup software must be assessed on the actual PC. Game launchers, driver tools, updates and security work can explain a higher count without an open foreground window.

Verify the gaming result

Compare the same warmed-up Fortnite/Valorant scenario before and after: identical resolution, rendering mode, driver, Reflex selection, input device and capture setup. Record frame times, stutters, average FPS and 1% lows, with supported latency measurements when available. Run several comparable trials. A smaller counter is not the acceptance criterion if the game feels worse.

NVIDIA's latency guide recommends Reflex where supported and says advanced system tweaks can help or hurt depending on configuration. This profile adds no undocumented timer, affinity or interrupt settings. The existing latency research explains the comparisons already included.

Recovery and verification status

Use Control Center 10 to restore all backed-up settings, then restart. Printing/indexing can be restored separately with 8. Removed optional apps may need Store reinstallation. Installed Brave and Discord remain installed.

Pure/mocked logic checks cover the service guards, original-state preservation, templates, hardware exclusions and sampled audit. Archive/manifest checks verify the delivered files. Windows installation, peripheral/game compatibility, FPS/input latency and the 50–70 process target have not been verified. Future-update compatibility and a continuous process maximum are not guaranteed.