Benchmarks
We measure things.
Or we don’t claim them.
The optimization scene runs on unverifiable “+50% FPS!!” screenshots. This page exists to do the opposite: a running, honest record of what our sessions actually change — captured properly, published with the boring details.
Status
First benchmark run — in progress
The first before/after runs are being captured now. They land here, in this shape, with the full method below — no cherry-picked single frames.
Method
How we’ll prove the work
Same machine, same everything
A benchmark only means something if the only variable is our work. Same hardware, same drivers version, same game settings, same scene — before and after.
Captured, not eyeballed
Frame data recorded with CapFrameX or equivalent capture: average FPS, 1% lows and frame-time consistency — not a screenshot of an FPS counter.
Multiple runs
Each scenario runs more than once. One lucky spike proves nothing; a consistent delta does.
Published with the raw story
Gains, no-gains and the hardware context get published together. A result we can’t reproduce doesn’t get published.
What gets measured
The numbers that matter,
not just the average
Average FPS flatters a stuttering game. These are the metrics that describe how a game actually feels.
Titles on the bench
Benchmarks cover the flagship competitive titles: Valorant, Fortnite, CS2, Call of Duty, Apex Legends. More titles follow — request yours in Discord.
Want your rig in the first batch?
We’re selecting pilot machines for the first published before/after runs. If yours is a typical build, you get a session and your results — good or average — on the record.