Docs/Using Frametide
Reading the results
Results compares every tested preset with your original settings, per game. A difference only counts when it's bigger than the normal ups and downs between your runs.
The headline#
- Winner: <preset>, with the change in average FPS and in the worst moments. Use this preset makes it active.
- No preset clearly beat your original settings. The differences were within the variation between runs. That's a real answer: your setup is already good for this game, and you've saved yourself tweaks that only feel faster. Try more runs, a steadier test scene, or a new round of presets with Improve on my test results.
- Test your original settings too: there's nothing to compare against yet.
What matters most? at the top changes how the presets are weighed: A bit of everything, Most FPS, Fastest response or Fewest stutters. The winner can change with it, the measured numbers don't.
The table#
| Column | Meaning |
|---|---|
| Average FPS | Frames per second on average. With 2 or more runs it shows ± the spread between runs, e.g. 448.2 ± 6.1. The bar compares presets at a glance. |
| 1% low | FPS during the slowest 1% of frames: the worst moments. Higher means fewer drops. Often the number you feel most. |
| 0.1% low | FPS during the slowest 0.1% of frames: the big hitches. |
| Stutters/min | Frames that took more than twice as long as usual (and at least 4 ms longer): visible hitches per minute. Lower is better. |
| Compared to original | Change in average FPS and lows, and the verdict: Better, Worse or No real difference. |
Hover a row for more: number of runs, how much they varied, CPU and GPU load, and notes. GPU load near 100% means the graphics card is the limit; lower usually means the CPU or the game engine is.
When runs vary a lot#
If the runs of one preset differ by more than 5%, Results warns you, e.g. “The runs of Your original settings differed by ±19% from each other, so smaller differences can't be told apart.” Usually one run went wrong: a loading screen, an alt-tab, a different part of the map. Check All runs, leave the odd one out, or test again following these tips.
The frame-time graph#
Under Frame times you see one run from start to end:
- The lime line is the average frame time. Lower is smoother; a flat line is good.
- Faint spikes are the slowest frame in each moment: the hitches you feel.
- The red dashed line is the stutter threshold. Spikes above it count as stutters.
- The grey line behind it is the newest run of your original settings, for comparison.
- Left axis: milliseconds per frame and the matching FPS (e.g. 2.5 ms = 400 FPS).
Click Graph next to any run under All runs to show it. Runs measured with app versions before the graph existed have no graph.
All runs#
Every measurement of the game, newest first, with date, preset, average FPS and 1% low. Untick a run that went wrong: it stays saved but no longer counts, and the table updates right away. With 3 or more runs of a preset, runs far away from the others are marked Unusual. Runs measured before a needed restart are marked as such.
Other notes you may see#
- Runs were made with different graphics drivers: driver updates can change performance by themselves. Retest the original settings for a fair comparison.
- Your original settings were last measured on <date>: more than 30 days before your newest test. Tick them in your next test.
- ⚠ next to a preset: some runs were measured without a needed Windows restart.
errors.log (Settings → Open data folder → logs).