Browser presentation diagnostic
Refresh rate test
Measure the timing cadence delivered to this browser tab, then inspect consistency and delayed callbacks without installing software.
- Real frame timestamps
- Bounded sample
- No permission
How to use
- 01 · PrepareMove this window onto the display you want to check. Keep it visible, focused, and still.
- 02 · MeasureChoose a length and press Start test. The first 0.5 seconds warm up and are not counted.
- 03 · ConfirmReview Hz and stability, then repeat while the computer is idle to check consistency.
Measurement
Warm-up + timed sample
Ready
Ready. No measurement runs until you press Start test.
Run analysis
Robust statistics from this run
Estimated browser presentation cadence, not a direct panel reading
- Median interval
- —
- Stability
- —
- MAD jitter
- —
- 95th percentile
- —
- 10% trimmed mean
- —
- Delayed callbacks
- —
- Interval range
- —
- Sample size
- —
Complete a run to get a stability and jank interpretation. Interrupted samples are never reported as results.
How the calculation works
- After the 0.5-second warm-up, sampling ends at the selected time or 6,000 intervals, with a 12,000-callback safety ceiling.
- Only positive gaps between consecutive requestAnimationFrame(timestamp) values are sampled.
- Hz = 1000 ÷ median frame interval (ms). The median resists occasional long stalls better than a plain average.
- MAD jitter is the median absolute deviation divided by the median. P95 exposes the slow end of frame delivery.
- A callback is marked delayed when its interval exceeds both a robust noise allowance and 1.5× the median. That signals browser timing disruption, not proven physical frame loss.
- “Very steady” requires MAD jitter ≤1.5%, delayed callbacks ≤0.5%, and first-to-last segment spread ≤2%. “Mostly steady” allows ≤4%, ≤2%, and ≤5%; anything wider is “Uneven.”
Important limit: this page observes browser presentation timing. Compositing, system load, power saving, variable refresh rate, browser caps, remote sessions, and which display the window is routed to can all affect it. It cannot certify a panel’s physical scan rate, cable mode, pixel response, or input lag.
Optional motion comparison
Compare browser-cadence motion with an intentionally simulated 30 fps reference at the same speed. This is a visual aid, not another measurement.
Off by default. The comparison also stops if this page loses focus.
No browser or device permission is requested. Measurements stay in this page and are not saved by this tool.