Slow-frame percentile analyzer

Ready
FPS over time · one point per second
Avg FPS —Min FPS —Max FPS —

Live test results

Average FPS— fps
Median FPS— fps
1% low— fps
0.1% low— fps
P95 frame time— ms
P99 frame time— ms
P99.9 frame time— ms
Worst frame time— ms

1% low needs at least 100 samples; 0.1% low and P99.9 need at least 1,000.

Ready

Slow-frame analysis

1% Low FPS Test

Run a 1% low FPS test to summarize the slowest browser frame intervals. Average FPS shows overall cadence; lows and high frame-time percentiles expose the brief delays that averages smooth away.

What Is 1% Low FPS?

1% low FPS is a summary of the slowest one percent of sampled performance. This page sorts valid browser frame intervals from slowest to fastest, takes the slowest one-percent subset, calculates its mean interval, and converts that mean with 1000 / milliseconds. The result is an equivalent FPS rate for the slow tail. A lower value than the overall average indicates that some frames took materially longer than typical.

The phrase “1% low” is used in game benchmarking, but this online implementation measures only browser-observed animation cadence. It cannot read frame telemetry from another process. The concept remains useful because it teaches how brief slow periods change perceived consistency and lets you compare browser sessions under controlled conditions.

1% Low vs Average FPS

Average FPS is calculated from mean interval across the full valid run. Most normal frames dominate that number, so a small group of delays may have limited impact. The 1% low deliberately isolates the slow tail. Median FPS represents the typical interval, while worst frame time represents one extreme. Used together, average, median, 1% low, P99 frame time, and worst time describe the center, tail, and single largest delay.

A large gap between average FPS and 1% low suggests uneven delivery, but it does not reveal the cause. A single system interruption, repeated long tasks, browser throttling, or page workload can contribute. Repeat the 1% low FPS test and inspect the highlighted slow portion of the timeline before concluding that the pattern is persistent.

How to Get Started

For 1% low, sample size matters more than it does for a current counter. Let the continuous test run for at least 30 seconds for a useful comparison and around 60 seconds or longer when you need enough samples for 0.1% low at a high refresh cadence. Keep the page visible and do not interact with unrelated tabs during the test.

  1. Set the Stress Level, press Start Test, and allow the warm-up to finish before reading the live values.
  2. Wait for at least 100 intervals before expecting 1% low and 1,000 before expecting 0.1% low or P99.9.
  3. Compare runs with the same duration, active display, browser state, and power mode.

How FPS Tester Calculates FPS Lows

Every positive interval below one second is eligible after warm-up. For 1% low, the tool requires 100 valid samples, selects ceil(sample count × 0.01) slowest intervals, averages them, and takes the reciprocal. The use of ceil ensures at least one interval contributes. The 0.1% low repeats the process with a 0.001 fraction and a minimum of 1,000 samples. An explicit insufficient-samples state is more honest than extrapolating from too little data.

P95, P99, and P99.9 frame time use linearly interpolated positions in the sorted interval list. A high frame-time percentile corresponds to a low-FPS event because longer milliseconds mean slower delivery. P99.9 is suppressed until 1,000 samples exist. Rounding happens only for display; calculations retain the browser timestamp precision available to JavaScript.

Why Low-FPS Moments Matter

Perception is sensitive to changes in cadence. At 60 FPS, a normal frame budget is about 16.67 milliseconds. A 50-millisecond callback occupies roughly three such budgets and can appear as a hitch even when surrounding frames are timely. Average FPS spreads that delay across the full run. A low-FPS statistic gives the slow tail enough weight to remain visible.

Lows are not automatically “dropped frames.” The browser API reports callback timing, not a certified presentation history. A long interval means the page did not receive its next observed animation timestamp on the expected cadence. The display, compositor, operating system, and browser may all influence that path. Careful language preserves what the sample actually proves.

When to Use a 1% Low FPS Test

Use the low FPS test when two browser sessions have similar averages but feel different, after changing performance or power settings, or while investigating occasional web-animation stalls. It can also help learners connect percentile language with a visible frame-time graph. When measuring a native game, use its own overlay or a presentation-capture tool and confirm that tool’s specific 1% low formula, because definitions can vary.

For reproducibility, record run duration and keep the same browser, display, window position, and foreground state. Repeat at least three times if a decision depends on a small difference. A long one-off interruption can change tail metrics sharply; repetition reveals whether it is typical or accidental.

What 0.1% Low Means and Key Limits

The 0.1% low focuses on one interval in every thousand, approximately. It is more sensitive to rare events and therefore needs much more data. Even with enough samples, one background interruption can dominate it. Treat it as a description of the captured session, not a universal prediction. P99.9 frame time offers the same tail from the interval direction.

Browser lows are not native-game lows. Tab visibility, browser policy, system work, active monitor selection, battery saver, and timestamp behavior affect results. This test stops when hidden, uses a new warm-up on the next start, and rejects invalid samples, but it cannot remove every external influence. It cannot diagnose a component or guarantee that an application will remain hitch-free.