CPU Benchmark - Test Your Processor

What Is This CPU Benchmark?

This free browser CPU benchmark measures how quickly your current browser executes a consistent combination of JavaScript integer arithmetic, bit operations and floating-point calculations. Each timed trial runs inside a dedicated Web Worker rather than on the page's main thread, so the interface remains usable while the test runs. Results reflect measured browser computation throughput, not the processor's verified model, clock speed, power consumption or native-app benchmark rating.

You can run a single-worker baseline or compare it with two to four workers running concurrently. Parallel speedup varies with the JavaScript engine, browser scheduling, CPU allocation, operating system and background activity; it is not proof of the number of physical cores.

How to Use This Benchmark

1. Use a plugged-in device or a consistent power setting for fair repeat runs. Close heavy background apps, keep this tab visible and allow the device to cool between extended tests.
2. Choose Single + parallel comparison to measure one worker and a simultaneous group, or choose Single worker only for a shorter baseline run.
3. Select the measured time per trial (0.75, 1.2 or 2 seconds), choose 3 or 5 repetitions and, if comparing parallel throughput, select 2–4 workers or Auto. More repetitions take longer and produce more heat.
4. Click Start Benchmark. Each trial first performs a short, unscored warm-up. Watch the live progress and trial table; press Stop to cancel or Reset to clear all results.
5. After completion, review the single-worker median, combined parallel median, measured speedup, trial variation and measured wall time. The Local browser index is a simple calculated convenience value, single-worker million iterations per second multiplied by 100, not an industry-standard score.
6. Use Copy Report or Download JSON to save your real measurements, configuration and testing limitations. For useful comparisons, repeat using the same browser version, worker settings and power conditions.

How It Works

One Web Worker executes batches of 4,096 iterations using a fixed JavaScript maths routine. Every trial starts with about 180 milliseconds of warm-up (excluded from scored work). The worker then repeats the same batch until the selected measured time has elapsed. performance.now() records the elapsed time inside that worker, while the program counts the iterations actually completed and returns a checksum to keep a real, observable computation result.

Throughput in million iterations per second (M iter/s) is the completed iteration count divided by measured seconds and one million. With multiple workers, the benchmark sums their iteration counts and divides by the longest measured worker interval. It reports the median across 3 or 5 trials to reduce reliance on a single noisy sample. Worker startup and warm-up are included in total wall time, not in the rate denominator.

The UI cancels a run if the tab becomes hidden and immediately terminates its active workers when you stop it. If workers cannot start due to browser restrictions, security policy or other errors, it displays an unavailable/error result rather than a fabricated score.

Understanding Your Results

Single-worker median is the observed throughput from one JavaScript worker. Parallel-worker median is the combined throughput from the selected number of simultaneous workers. Speedup is parallel median divided by single median, so it can be smaller or larger than one; it cannot be interpreted as physical CPU core count. The in-page bars only compare the two measured results within the same run.

The Local browser index is calculated as M iter/s × 100 for convenience. It is not directly comparable with Geekbench, Cinebench, hardware reviews or other sites' benchmark numbers. A browser's reported logical-core count may be reduced or capped for privacy, and a worker does not map reliably to one physical CPU core. The first-to-last change and min–max spread show variability but do not independently diagnose overheating or throttling.

Benefits

• Measures a consistent browser JavaScript workload without installations or external benchmark servers.
• Provides repeat-trial medians and real calculation counts rather than guessing hardware specifications.
• Lets you inspect how this browser handles one worker versus simultaneous worker tasks.
• Offers manual Stop, automatic background-tab cancellation and portable local JSON reports.

Common Mistakes to Avoid

• Comparing unlike settings: Duration, worker count, browser engine and version all affect results. Keep them the same when comparing runs.
• Treating scores as native CPU ratings: This measures browser JavaScript execution, not CPU frequency, operating voltage, physical cores, CPU utilization percentage or a calibrated native benchmark.
• Running too many tests back-to-back: Longer repeated workloads warm the CPU. Use a consistent power setting, take breaks and follow device safety guidance.
• Trusting one surprising run: Background tasks, extensions, garbage collection and operating-system scheduling can create outliers. Look at the median and trial spread.
• Assuming a parallel ratio proves core count: Worker scheduling, browser limits and workload saturation can all affect scaling.

Frequently Asked Questions

Why Browser Scores Differ From Native Benchmarks

JavaScript runs inside a browser runtime with its own just-in-time compilation, security sandbox and task scheduling. Browsers can reduce timing precision for privacy; power saving, thermal limits, memory pressure and other software also affect observations. Web Workers are useful for keeping the interface responsive and testing concurrent JavaScript work, but they do not provide low-level hardware counters, CPU temperatures or verified per-core affinity. Because browser engines are different, cross-browser scores must be interpreted cautiously.

Conclusion

This CPU Benchmark provides transparent, local measurements of a repeatable JavaScript workload with single-worker and optional parallel-worker trials. Use it to compare sessions under controlled conditions, not to certify processor specifications. Pair it with the GPU Tester and Network Latency Test for other kinds of browser diagnostics.