Windows gaming tweaks can sound convincing when an FPS counter jumps after several changes. To find out what helped your PC, isolate each adjustment, repeat the same scene and keep its trade-offs visible.
Can Windows 11 tweaks actually improve gaming performance?
A useful Windows change should solve an identified problem and survive a repeat test. Start by checking which GPU runs the game and whether unwanted software is consuming resources. Treat a higher FPS counter, steadier frame delivery and lower input latency as separate outcomes.
This guide is an evidence audit and a proposed test method. HD OPTI has not benchmarked these changes here, and there is no measured uplift to promise. Applying several tweaks together cannot tell you which setting helped.
Check the GPU before changing Windows services
On a PC with multiple GPUs, open Settings > System > Display > Graphics, select the game and open its graphics options. Choose High performance and check the GPU named beside it, save, then restart the game. Labels can vary with the Windows build. Record the original selection so you can restore it.
This preference selects a GPU for an application. It does not prove that a game previously used the slower GPU, and it should not be confused with the Windows power mode. Compare the actual GPU used before and after. If you corrected an unintended GPU assignment, describe the result as that correction rather than a general Windows speed boost.
Seven common changes: what deserves a test?
| Setting | Useful check | Decision for your first test |
|---|---|---|
| Startup apps | In Settings > Apps > Startup, review apps that launch automatically | Disable only an optional app you do not need; test after restarting |
| Game Bar | Record whether the overlay or recording features are actually in use | Compare the specific feature you use; avoid assuming one switch removes every capture workload |
| Graphics preference | Check the GPU assigned to the actual game executable | Correct an unwanted assignment before testing unrelated tweaks |
| Windows Game Mode | Record its current state | If comparing on and off, make this the only changed setting |
| Delivery Optimization | Check sharing settings and update activity | Treat sharing as a network question; do not assume an FPS gain |
| Third-party services | Identify a suspected software conflict | Use temporary isolation and restore normal operation afterwards |
| Memory Integrity | Record the current protection state | Keep it enabled for the initial tests |
Startup settings control automatic launch. A lower startup-app count is not an in-game performance measurement. Use Task Manager to look for CPU, memory or disk activity, then compare with the suspected optional application closed. Record exactly what changed.
The Process Count Myth: What Actually Slows Down Your Windows PC
Turning off peer sharing in Delivery Optimization still allows updates from Microsoft. Its Activity Monitor shows monthly transfer totals, so use those as context rather than proof that an upload occurred during a particular hitch. Leave bandwidth controls alone unless you have a specific network problem to investigate.
Hiding Microsoft services in System Configuration only filters the list. It does not establish that every remaining service is dispensable. A clean boot is a diagnostic environment that can temporarily remove functionality. If isolation helps, re-enable items systematically to find the cause, then restore your normal startup configuration.
Desktop visual effects are a separate comparison. Search Start for Adjust the appearance and performance of Windows and record the Visual Effects selection. Reducing effects can improve desktop responsiveness, but that does not establish a change to a game's rendering performance. Test this separately if it addresses your symptoms.
Keep Memory Integrity out of the first optimisation pass
Memory Integrity uses virtualisation-based security to protect kernel code integrity. Its performance cost depends partly on processor capabilities. Switching it off removes that protection; it is not a free performance upgrade.
You can inspect its state in Windows Security > Device security > Core isolation details > Memory integrity. Our recommendation is to keep it enabled while checking GPU selection and optional background workloads. Any later comparison needs a separate, deliberate security decision and measurements from your own configuration. This guide does not recommend disabling it as a routine gaming step.
A repeatable FPS test for Fortnite, VALORANT or CS2
Use a repeatable local or practice scenario where available. Test each game separately. The following run count and capture length are a suggested protocol, not benchmark results or mandatory tool defaults.
- Write down the Windows build, game version, GPU driver, CPU, GPU, RAM, resolution and graphics API. Keep the laptop plugged in with the same power profile and cooling conditions.
- Keep the map, route, graphics preset, upscaling, frame generation, FPS cap, V-Sync and display mode unchanged. Keep monitoring and capture software identical between conditions.
- Warm up the scene before recording. Choose a 60-second route with fixed start and end points, excluding loading screens and menus.
- Record at least three baseline runs as condition A. Save the raw captures, average FPS, 1% lows and frame-time plot using the same tool version and low-percentile calculation throughout.
- Change one setting for condition B. Restart the game, or Windows when the setting requires it, and repeat the same warm-up and three captures.
- Restore condition A and repeat it. Compare the median result and the spread of individual runs, not just the best capture.
PresentMon can capture frame data to CSV. Record the exact metric names and version you use. Its DisplayLatency measures from the start of CPU work on a frame to display; it is not interchangeable with a physical mouse-click-to-visible-response measurement. Only report an input-latency improvement when the measurement actually covers that path.
Keep generated-frame settings fixed and identify whether your FPS metric concerns application frames or displayed frames. Frame-type tracking can require application or driver instrumentation. If a metric is unavailable, record it as unavailable rather than estimating it from average FPS.
High FPS but Game Still Stutters? How to Fix Frame Time Spikes
Decide what the result actually supports
| Observation | Interpretation to investigate | Next action |
|---|---|---|
| Changed runs overlap the baseline spread | The test has not separated an effect from run variation | Repeat under tighter conditions; do not advertise an uplift |
| The improvement disappears when A is restored and returns with B | Stronger evidence of an effect in this scenario | Repeat in another scene before generalising |
| Average FPS rises but slow frames worsen | The metrics disagree about the benefit | Review the frame-time plot and choose based on your priorities |
| A large gain follows a GPU assignment change | The GPU correction is a candidate explanation | Verify the GPU used in each condition |
| Fewer processes but unchanged game captures | The count fell without a demonstrated game benefit | Keep the change only if it serves another useful purpose |
A useful record contains every run, the exact changed setting and any lost functionality. Keep a tweak only when its repeatable benefit addresses your problem and justifies its trade-off. If nothing separates from normal variation, “no demonstrated improvement in this test” is a useful result.
FAQ
No universal gain is established. On a system with multiple GPUs, the preference selects a GPU for the application. Check the named device and actual GPU use, then compare the same scene after restarting the game.
Start with an optional app linked to unwanted resource use. Startup controls affect automatic launch, while service isolation can remove functionality. Use temporary diagnostic changes and restore normal operation after finding the cause.
Disabling peer sharing still allows updates from Microsoft. It changes where update files come from and whether they are shared; it does not establish an FPS or ping improvement.
DisplayLatency starts at CPU work on a frame. A click-to-display measurement starts earlier in the input path. Report the metric you actually captured and do not substitute one for the other.
Keep it enabled for the initial optimisation checks. It protects kernel code integrity through virtualisation-based security, and disabling it removes that protection. This guide supplies no measured gain that would justify a blanket recommendation to switch it off.




