Benchmark Builder

About Benchmark Builder

Compare the speed of JavaScript snippets in your browser. Write a shared setup block and any number of named snippets; each runs in a Web Worker, first calibrated into batches and then timed with performance.now() for a time budget (250 ms to 5 s) or a fixed number of iterations. Results show operations per second, mean time per run, relative margin of error at 95% confidence, total runs and speed relative to the fastest snippet. Code runs locally via new Function, so only run code you trust.

Common uses

• Compare for, for...of and reduce on the same data
• Check whether a micro-optimization actually makes a difference
• Measure the cost of a regex, JSON.parse or string-building approach
• Share a quick performance comparison with a teammate

FAQ

How accurate are browser micro-benchmarks?

They are indicative. JIT compilation, garbage collection, dead-code elimination and CPU throttling all affect results, so differences under a few percent or within the margin of error are usually noise. Run several times and test on the browsers you target.

What does the plus-minus percentage mean?

It is the relative margin of error: 1.96 times the standard error of the sample means, divided by the mean. Smaller is more reliable.

Why does my snippet need to use its result?

If a computation's result is never used, the engine may optimize part of it away and report unrealistically fast times. Assigning or returning the result reduces that risk.

Can the benchmark access the page or DOM?

No. Snippets run inside a Web Worker, which has no DOM access, so DOM operations cannot be benchmarked here.