An FPS counter is useful, but one average cannot describe every pause or how quickly a game responds. A sensible comparison records the hardware, game version, resolution, preset and rendering features alongside the result. Without that context, two impressive numbers may be measuring different experiences.
Average FPS is the broad view
Average frame rate summarises a capture period. It helps compare overall throughput when the tests cover the same scene under the same conditions. It does not show where slow frames occurred. A loading pause inside one run can also make the comparison different from a run that excluded loading.
Frame time describes the time associated with each frame. As a simple conversion, evenly spaced 60 FPS corresponds to roughly 16.7 milliseconds per frame; 120 FPS corresponds to roughly 8.3 milliseconds. These are mathematical examples, not measurements of a particular game or PC.
Read the low figure’s definition
NVIDIA’s FrameView guide distinguishes its 1% Low FPS measure from an older 1% FPS figure. Its low measure averages the slowest one percent of frames, aiming to capture stuttering more clearly. That is a reason to read a tool’s documentation rather than assume every label called “1%” uses identical calculations.
Compare low results produced by the same tool, version and capture method. Include a frame-time graph where possible. One minimum value may represent an isolated interruption; a pattern of repeated long frames tells a different story. Neither figure identifies the cause on its own.
Label generated frames separately
Frame generation inserts additional generated images between rendered frames. NVIDIA pairs its DLSS frame-generation technologies with Reflex for responsiveness, but a displayed FPS figure with generation enabled should not be presented as though the game rendered every one of those frames traditionally.
Keep a baseline without frame generation, then report the generated-frame result with its settings clearly labelled. Look at visual stability and input response as well as the output counter. NVIDIA’s developer guidance warns that some high-multiplier and V-Sync combinations on low-refresh displays can substantially increase input latency.
Make a repeatable comparison
- Choose a fixed benchmark or a route you can repeat.
- Use the same capture duration and exclude loading consistently.
- Record native resolution, upscaling mode and frame-generation state.
- Repeat the run and investigate large variation before publishing.
- State the measurement tool and which FPS definition you used.
Our buying guides should identify independent test-system results and estimates clearly. A graphics-card chart from another PC cannot guarantee the same average or low on a different CPU, memory configuration or game update. Use the measurements to understand trade-offs, then retain the conditions that make them meaningful.
Sources & further reading
Checked 17 Nov 2026. Source-based reporting and guidance; individual test results are identified where available.
Tell us the article title, the detail and a supporting source.
Corrections & contact →