Average FPS sums up how many frames fit in one second, but it doesn’t show how those frames reach the screen. A session can have a high average and still show noticeable stutters at its worst moments.
Frametime is how long each frame takes, in milliseconds. The more even the series, the more predictable the response. 1% low and 0.1% low summarize the slow tail of the distribution; they’re signs of consistency, not standalone proof of a cause.
To compare a tweak, keep the game, resolution, FPS cap, refresh rate, power settings, driver, map and capture path the same. Warm up, repeat the session and log the environment. If a capture is missing the process identity, frame data or fresh telemetry, throw it out.
Rise Test applies this protocol and keeps measured results apart from hypotheses. A verdict can be improved, neutral, worse or inconclusive. Whether you keep a tweak comes from the series of sessions, not from a nice-looking bar or a one-size-fits-all promise.
