LeCoreOS 11 network research and setup — October 15, 2026

What can reduce which delay

The package has not measured your ping, packet loss or loaded latency. No millisecond reduction is promised. A PC registry file cannot configure a router's queues. Use the right measurement for each symptom; Fortnite input feel alone does not identify a network cause.

First establish a baseline

1. Use a direct Ethernet connection to your router for the first test. If Wi-Fi is required, compare a clear supported band and a stable signal; radio controls are in PC settings. Select the nearest suitable server region in each game.
2. Pause game/client updates, cloud sync and streaming temporarily. Save work before closing apps. Keep required audio/controller/capture tools and services available.
3. Open Bufferbloat test or Cloudflare speed test in the Network page. These third-party tests use data and send traffic to their test servers. Record idle RTT, loaded download/upload RTT, loss/jitter where available, and sustained download/upload Mbps. Repeat several times at relevant busy hours.
4. Compare game RTT/loss using the game's own telemetry. A browser test endpoint can use a different route from the game server, so its RTT is not the game's RTT.
5. Repeat with the household's normal traffic and required streaming/capture workloads. A dual-PC stream can still saturate a shared uplink.

Address bufferbloat at the bottleneck

OpenWrt's SQM guidance describes combining traffic shaping, scheduling and active queue management to control latency while the connection is busy. A router with CAKE or fq_codel-based SQM, and enough CPU for your link rate, can put a controlled bottleneck before an oversized modem/ISP queue. Marketing labels such as Gaming QoS or device priority do not establish which queue management is implemented.

1. Confirm your existing router supports an SQM implementation through its vendor documentation. The app does not install router firmware. Keep an independent backup of router configuration before making router changes.
2. Enter measured stable download and upload Mbps in LeCoreOS 11, not the advertised tariff or the NIC's link speed. Apply / save SQM recommendation saves 90% starting rates in kbit/s. Example: measured 500/80 Mbps gives 450000/72000 kbit/s. The saved file explicitly says RouterConfigured=false.
3. In the router's documented SQM interface, select the actual WAN/bottleneck interface and CAKE or an appropriate fq_codel setup. Enter the saved down/up shaping rates. Account for link-layer overhead according to the router/ISP link documentation; do not guess a universal Ethernet/PPPoE/cable overhead value.
4. Apply at the router, then repeat the same idle/loaded tests. If loaded latency remains excessive, lower one relevant rate and retest. If capacity is unnecessarily lost, cautiously increase it while checking loaded latency. 90% is a starting point, not an optimum or a guarantee. Variable cellular/cable rates may require a more conservative or adaptive setup.
5. Check router CPU utilization. SQM can become CPU-bound at high rates. Hardware flow/offload features can bypass or conflict with a particular SQM setup; follow the router's own compatibility guidance rather than changing all acceleration blindly.
6. Confirm ordinary connectivity, streaming and games remain stable. Keep the measured better setup or use the router's saved configuration to revert. LeCoreOS 11's Windows recovery button cannot undo a manually changed router.

The app does not reserve bandwidth, install a traffic driver or set an arbitrary Windows-wide throttle. Its saved plan and browser test buttons are setup aids. When the link is idle, SQM may have little effect on propagation/routing delay.

Supported Ethernet comparisons

Choose your connected physical 802.3 Ethernet adapter in the app. Wi-Fi, VPNs, virtual switches, Bluetooth and capture interfaces are excluded. Only existing standardized driver properties with advertised valid values are writable.

Apply one change, restart Windows, then repeat the same test and game scene. The app uses -NoRestart so it does not drop your connection during the action; many driver changes need restart before taking effect. Orange reflects configured readback, not active interrupt behavior, actual RSS distribution, measured ping or SQM operation. Unsupported controls are disabled rather than inventing a registry value.

Original values are durably recorded in C:\ProgramData\LeGameOS\Network.json before writes. Repeated toggles retain first originals. Full machine recovery and the separate Network recovery controls include these entries. If hardware or driver identity changed, recovery reports the mismatch and keeps the original for manual inspection in Device Manager. Do not delete this file to suppress a warning.

Microsoft's NIC tuning article is for Windows Server and discusses hardware/workload-dependent tradeoffs. Only its supported concepts and driver APIs are used here. Server-specific BIOS, interrupt affinity and fixed CPU tuning are not copied into a universal gaming profile.

Changes deliberately not applied

DNS changes do not reduce the distance/route of an already connected match. TCP ACK/Nagle recipes are not a universal fix for UDP game traffic or click-to-display latency. Disabling IPv6, changing MTU without path evidence, disabling every checksum/segmentation offload, forcing duplex, arbitrary DSCP policies or replacing a congestion-control algorithm can impair another workload. Those settings remain available through their supported tools when a specific measured problem calls for them; this playbook does not blanket-write them.

Defender, Windows Firewall, network connectivity, DHCP/DNS client services, game anti-cheat, controllers and audio dependencies stay available. No network service is killed to meet a process-count target.

Primary references

• OpenWrt SQM overview and setup
• OpenWrt SQM details, shaping rates and bottleneck placement
• Microsoft NIC performance tradeoffs (Server context)
• Microsoft interrupt moderation
• Microsoft RSS introduction
• Microsoft standardized enum values
• Microsoft Get-NetAdapterAdvancedProperty and advertised values
• Microsoft Set-NetAdapterAdvancedProperty and NoRestart
• Microsoft physical adapter identity/media properties
• Cloudflare Speed Test
• Waveform's bufferbloat measurement tool

Sources support the concepts and interfaces. They are not a benchmark of this release or your PC/router. Windows device behavior and gaming performance still require testing on your hardware.