How to Use JetStream 3 for Modern Web Performance Testing

Introduction

JetStream 3 is the latest cross-browser benchmark suite developed collaboratively by the WebKit, Google, and Mozilla teams. Unlike earlier versions, JetStream 3 focuses on real-world web application scenarios, especially the evolving role of WebAssembly. This guide walks you through how to use JetStream 3 to measure browser performance accurately, interpret its scores, and apply the insights to your development workflow.

How to Use JetStream 3 for Modern Web Performance Testing — Source: webkit.org

What You Need

A modern web browser (Chrome, Firefox, Safari, or Edge) that supports WebAssembly and JavaScript ES6+
A stable internet connection (the benchmark runs entirely in the browser, but some assets may load from CDN)
A device with at least 4GB RAM and a reasonably modern CPU; results are more consistent on desktop/laptop
Patience: a full run of JetStream 3 takes about 15–25 minutes
Optional: a secondary browser for comparison testing

Step-by-Step Guide

Step 1: Access the JetStream 3 Benchmark

Open your preferred browser and navigate to the official JetStream 3 website (usually hosted at jetstream.benchmarks.com or a similar URL provided by the team). Ensure you are on a stable network and close unnecessary tabs to minimize interference.

Step 2: Start a Standard Benchmark Run

On the homepage, click the large “Run Benchmarks” button. The suite will automatically begin running a series of tests covering JavaScript, WebAssembly, and mixed workloads. Do not switch tabs or interact with the browser during the run, as user activity can skew results.

Step 3: Understand the Scoring and Phases

JetStream 3 divides its tests into categories. For WebAssembly, it now uses an integrated metric rather than separate “Startup” and “Runtime” phases (as JetStream 2 did). The score for each test is computed as Score = 5000 / time (in ms). If a test completes in under 1 ms, the benchmark clamps the score to 5000 to avoid an infinite score—a lesson learned from JetStream 2 where zero‑ms startup times caused division by zero.

Step 4: Watch for the WebAssembly Infinity Problem

As browser engines optimized WebAssembly instantiation, some micro-benchmarks reached sub‑1 ms times. In JetStream 2, Date.now() rounded such times to 0 ms, producing infinite scores. JetStream 3 addresses this by using higher‑precision timing and clamping scores. During your run, if you see any test scoring exactly 5000, it means that test ran extremely fast (under 1 ms). This indicates that the workload is no longer challenging for modern engines—a sign to focus on more complex tasks.

Step 5: Interpret the Overall Score

After completion, JetStream 3 displays a composite score (e.g., 1500 points). Higher is better. The composite is a geometric mean of all individual test scores. Pay attention to sub‑scores for JavaScript and WebAssembly separately, as they reveal where a browser excels or lags. For WebAssembly, especially look at tests that involve large C++ codebases (like image decoders) and UI frameworks—these reflect real‑world usage.

Step 6: Compare Across Browsers (Optional)

To compare performance, run the same test on different browsers on the same machine. Use incognito/private modes to avoid extension interference. Take at least three runs per browser and average the results. JetStream 3 is designed for cross‑browser consistency, so differences reflect engine efficiency, not random noise.

Step 7: Apply Findings to Your Development

If you are a web developer, use JetStream 3 results to decide which browser engine to target for performance‑critical features. For example, if your app uses heavy WebAssembly (e.g., video editing or image processing), a browser with a high Wasm score will deliver faster startup and smoother interaction. Track benchmark trends over time to see how engine updates affect your use case.

Tips for Accurate and Useful Benchmarking

Close background apps: Other software (especially video streaming or cloud syncing) can increase system load and skew results.
Run multiple times: One run may be affected by thermal throttling or background updates. Three runs with a cooldown period between them provide a reliable average.
Ignore absolute numbers across devices: A score of 5000 on a low‑end phone vs. a high‑end laptop are not directly comparable. Focus on relative performance within the same device.
Beware of the “infinity trap” in older benchmarks: If you still use JetStream 2, remember that sub‑1 ms Wasm instantiation times can produce artificially inflated scores. JetStream 3 fixes this, but always note the version when citing results.
Update your browser: The latest engine versions usually include recent optimizations for Wasm and JavaScript. Always test with current stable releases.
Use the integrated developer tools: Browsers like Safari and Chrome offer performance profilers. Combine benchmark scores with profiling data to pinpoint specific bottlenecks (e.g., call stack overhead, memory allocation).

By following these steps, you can leverage JetStream 3 to gain meaningful insight into modern web performance. This suite is invaluable for both browser engineers optimizing their engines and web developers aiming to deliver fast, responsive applications.

Tags: