CPU Benchmark
Measure browser compute performance with an integer workload and up to 64 threads to evaluate your hardware capabilities. Runs in your browser, GDPR-compliant.
Your inputs are processed in your browser and are not transmitted to our servers. Note: third-party resources (e.g. advertising and analytics from Google/Cloudflare) and an optional PayPal donation link may transfer data when loading or when clicked. Browser extensions or plugins may be able to read content that is visible in the input fields.
The result will appear here …
How to use this tool (video)
This video is hosted on YouTube. When you play it, data may be sent to Google.
CPU Benchmark in your browser: measure compute performance
This CPU benchmark runs a fixed integer workload in your browser. Choose from one logical thread up to the number reported by your browser, with a limit of 64. Each worker receives the same amount of work. The result includes elapsed time, operations per second, and a normalized score.
How the test works
A short warm-up runs before every measurement. The selected Web Workers then execute the same deterministic loop. Each worker returns its operation count and checksum, and the browser combines the results. The test uses standard Web APIs, so it works on x86 and ARM devices. Nothing about your measurement is transmitted to our server.
The score uses a simple reference: one point equals 1,000,000 operations per second. Compare scores only with the same benchmark version, workload, thread count, and browser settings. The current benchmark version is 2.3.0. The scalar CPU mode uses WebAssembly without the SIMD extension. A separate gzip compression mode uses the Compression Streams API. Each round repeatedly compresses a fixed 1 MiB block so the test does not finish in a few milliseconds. Its values are a different workload and are not directly comparable with the CPU score.
Why results can vary
- Thermal throttling can start when the device is hot or running on battery.
- Browser versions, power modes, and the number of available threads affect the result.
- Other tabs, extensions, and running applications create background load and compete for CPU time.
- A browser benchmark is not a universal hardware score and does not replace native application tests.
For a fair comparison
Close unnecessary tabs, use the same browser, and repeat the test under similar conditions. The long workload takes more time and can make variation easier to see. Use the standard workload for a quick run.
Frequently asked questions
What does this benchmark measure?
The benchmark runs a fixed integer workload with Web Workers in your browser and measures elapsed time, operations per second and a normalized score. One point equals 1,000,000 operations per second.
How do I choose the number of threads?
You choose between one logical thread and as many as your browser reports, up to 64. Each worker receives the same amount of work so the load is distributed fairly.
Why do my results vary between runs?
That is normal. CPU throttling, browser settings, power modes and background load from other tabs or programs influence the result. For a fair comparison close unnecessary tabs and repeat the test under similar conditions.
What is the gzip mode?
The separate gzip mode repeatedly compresses a fixed 1 MiB block through the Compression Streams API. Its values measure a different workload and are not directly comparable with the CPU score.
Can this result be compared with other devices?
Only to a degree. Compare values only with the same benchmark version, workload, thread count and browser settings. A browser benchmark does not replace a measurement with native applications.
Are data uploaded during the measurement?
No. Everything runs entirely in your browser. No data is transferred to any server.