Benchmark Microsoft Flight Simulator on PC with a repeatable test, built-in FPS display, frame-time checks and clear CPU-versus-GPU diagnosis.
To benchmark Microsoft Flight Simulator on PC, create a repeatable flight, lock the resolution, graphics, weather, traffic and aircraft, then measure FPS and frame time with Developer Mode’s Display FPS overlay. Run the same scene several times, discard the first streaming-heavy pass, and compare the median results—not a single peak frame rate.
This method works for Microsoft Flight Simulator 2020 and 2024. Menu wording may differ between them, while MSFS 2024’s heavier dependence on streamed content makes connection and cache conditions especially important.
How to run a repeatable MSFS benchmark
- Establish a stable baseline. Restart the PC, close unnecessary background applications and finish any simulator or system updates. Record the simulator build, graphics driver, CPU, GPU, RAM and display resolution. Start a new comparison after any simulator or driver update rather than mixing results.
- Lock every display setting. Record the output resolution, render scaling, graphics preset, anti-aliasing or upscaling mode, DirectX mode, V-Sync, FPS cap and frame generation state. A preset labelled Custom cannot be compared reliably without listing its individual changes. Our practical MSFS graphics-setting guide explains which controls need to remain fixed during testing.
- Remove changing conditions. Select a weather preset and fixed time, then disable live weather, multiplayer and live or AI traffic unless one of those features is the subject of the test. Use the same airport, runway or gate, aircraft, livery, fuel load and cockpit state.
- Fix the camera. Cockpit and external views produce different loads. Glass displays, cabin interiors, reflections and field of view all affect performance, so save or carefully reproduce one view and do not pan the camera during the measured section.
- Open the performance display. Enable Developer Mode in the simulator settings, open the developer toolbar’s Options menu and select Display FPS. This provides frame-rate, frame-time and bottleneck information such as
Limited by MainThreadorLimited by GPU. - Warm the scene before recording. Let the aircraft and scenery settle for roughly two minutes. Discard the first pass if it includes loading, shader compilation or heavy scenery streaming, then capture at least three runs of about 60 seconds each.
- Change one variable at a time. Adjust one setting, restart the scenario and repeat the same procedure. Changing render scaling, traffic and Terrain Level of Detail together may improve FPS, but it will not reveal which setting caused the change.
What test scene should you use?
A useful benchmark suite contains at least one demanding airport test and one airborne GPU test. One scene cannot represent every phase of an MSFS flight.
| Test scene | What it exposes | How to keep it repeatable |
|---|---|---|
| Parked at a detailed airport | Main-thread load, airport detail, traffic and cockpit systems | Use a fixed gate, stock weather, parked aircraft and an unchanged cockpit view |
| Short final into a large airport | Scenery loading, traffic, aircraft systems and landing-related stutters | Start from the same position and altitude with identical weather and traffic |
| Cruise above cloud | GPU load from resolution, clouds, anti-aliasing and reflections | Hold the same altitude, heading, weather preset and camera |
| Low flight over streamed scenery | Frame-time spikes, storage and scenery delivery | Use the same route and distinguish a cold first pass from warmed runs |
Use a default aircraft and unmodified airport when comparing different PCs. Use your favourite complex aircraft, add-on airport and normal traffic settings when tuning your own system; that result is more personally useful, but it cannot be compared fairly with somebody else’s stock benchmark.
Which benchmark metrics should you record?
Average FPS alone hides the pauses that make a flight feel rough, so record consistency and the reported bottleneck as well.
- Average FPS: use the median result from repeated runs rather than the best run.
- Frame time: lower and steadier is better. For reference, 60 FPS corresponds to about 16.7 milliseconds per frame, while 30 FPS is about 33.3 milliseconds.
- One-percent low FPS: if your chosen capture tool reports it, this is more useful than an absolute minimum caused by one loading event.
- Frame-time spikes: record visible pauses even when average FPS remains high. Our frame-time and stutter troubleshooting covers cache, add-on and FPS-cap checks.
- Limiter status: note whether the developer display reports MainThread, GPU or alternating limits.
- Visual quality: include screenshots or notes about cloud quality, object pop-in, glass-cockpit clarity and upscaling artefacts. A faster result is not equivalent if image quality changed.
How can you tell whether MSFS is CPU- or GPU-limited?
Limited by MainThread usually means the simulator’s critical CPU-side work is determining performance, while Limited by GPU means graphics rendering is taking longer.
- MainThread limit: test lower Terrain Level of Detail, Object Level of Detail, traffic density and airport activity. Complex aircraft systems and detailed scenery can also increase CPU-side load. Overall CPU utilisation may still look low because one critical thread can be saturated while other cores are not.
- GPU limit: test lower render scaling, output resolution, cloud quality, shadows, reflections and ambient occlusion. Watch VRAM use because exceeding practical video-memory capacity can cause severe spikes rather than a simple gradual FPS loss.
- Alternating limit: the workload is balanced or changing with the view. Repeat the test before drawing a conclusion, especially if the difference between frame times is small.
Once the limiter is clear, use our CPU-versus-GPU performance diagnosis to target the settings that can actually help. Lowering resolution rarely fixes a strong MainThread limit, while reducing traffic will do little for a pure GPU limit.
Should frame generation be enabled during an MSFS benchmark?
Benchmark first with frame generation disabled, then run a separate test with it enabled if that is how you normally fly. The first result shows the underlying rendered performance; the second shows the presented experience.
Do not compare generated FPS with another result that has frame generation disabled. Some counters report rendered frames and others report displayed frames, so use the same overlay throughout and label the result clearly. Frame generation can raise displayed FPS, but it does not remove a weak base frame rate, main-thread pauses or slow control response.
Why do repeated benchmark results vary?
Most unexplained variation comes from streaming, changing traffic, camera movement or background activity rather than the graphics setting being tested.
- The first run may compile shaders or fetch scenery that later runs reuse.
- Live weather, multiplayer and traffic produce different workloads on every pass.
- MSFS 2024 can vary more when connection quality or streamed-content availability changes.
- Saved flights may not restore every aircraft system or third-party add-on state consistently.
- A simulator, aircraft or scenery update invalidates strict comparisons with older runs.
- Thermal throttling can make later tests slower, so monitor temperatures and keep the same cooling conditions.
When publishing or comparing a score, include the hardware, simulator version, scenario, native and output resolution, render scaling, preset, frame generation state, average FPS and one-percent low. Compare the result against practical MSFS CPU, GPU, memory and storage expectations, not against an unexplained FPS number from a different aircraft or airport.