RadiantOPTIMIZER GUIDE

TESTING NOTEBOOK · 02 / COMPARE

Compare your FPS runs,
without filling in the gaps.

Enter the values from matching repeated runs. This tool calculates simple summaries of what you entered; it does not scan your PC, collect a benchmark, or predict a result.

Method note updated

YOUR RECORDED RUNS

Before / after summary

Use 3–10 positive decimal values from 0.001 to 1,000,000 FPS in each Average FPS field. Separate runs with commas, spaces, or new lines; commas are separators, not thousands markers.

Before change

Keep game build, scene, resolution, quality settings, and capture method matched.

Example: 62.4, 61.8, 63.1. Allowed range: 0.001–1,000,000 FPS.

Use the 1% low reported for each run. This tool averages those reports; it does not recalculate a percentile.

After change

The number of runs can differ from Before, but conditions should remain matched.

Example: 66.7, 67.2, 65.9. Allowed range: 0.001–1,000,000 FPS.

If you add 1% low values, add both groups, match each group’s Average FPS run count, and keep each low value at or below its corresponding average.

HOW TO USE THIS

Keep the comparison honest.

  1. 01
    Record repeated runs

    Capture 3–10 Average FPS values for the same repeatable scene before making a change.

  2. 02
    Match the conditions

    Game build, drivers, settings, resolution, capture method, and background load can all affect a comparison.

  3. 03
    Read the range with the mean

    Averages summarize the entries. Overlapping run ranges call for caution and more matching runs.

METHOD

A transparent calculation

For each group, the tool reports the arithmetic mean, minimum, maximum, run count, and percent change in mean from Before to After. A range is the lowest through highest entered run.

Read the full testing methodology before comparing changes, then keep raw notes in the testing worksheet.

LIMITS

What this cannot establish

This is not a frame-time capture or statistical significance tool. It cannot establish causation, guarantee an improvement, or say how a change will behave on another PC.

1% low values are the values your measurement tool reported for each run. They are averaged here as entered, not reconstructed from frame-time samples.