In a landmark collaboration with Google and Mozilla, the WebKit team has unveiled JetStream 3.0—a major overhaul of the cross-browser benchmark suite. This update isn't just a tweak; it's a fundamental rethink of how we measure browser performance, especially for modern web applications and WebAssembly (Wasm). While the joint announcement highlights the collective effort, this article dives into the key innovations, from solving the infamous 'infinity problem' to rethinking Wasm startup metrics. Whether you're a developer or a performance enthusiast, these ten insights will show why JetStream 3 sets a new standard for benchmarking.
1. Why a New Benchmark Was Necessary
Benchmarks are essential for driving performance improvements, but they can become stale as web practices evolve. JetStream 2, released years ago, no longer reflected real-world usage—especially for WebAssembly, which has moved from niche startups to critical-path libraries. Moreover, when low-hanging optimizations are exhausted, engines risk overfitting to synthetic workloads. JetStream 3 was born to address these issues, refreshing both the test set and the scoring philosophy to match today's dynamic web.
2. WebAssembly Takes Center Stage
One of the most dramatic shifts in JetStream 3 is how it handles WebAssembly. In JetStream 2, Wasm benchmarks were split into distinct Startup and Runtime phases. This made sense when Wasm was primarily used for large C/C++ ports that tolerated long initialization. Today, Wasm appears in image decoders, UI frameworks, and even in page-load critical paths. JetStream 3 drops the artificial separation, modeling Wasm performance as an integrated part of the overall application flow.
3. The Infinity Problem—and Its Solution
Browsers became so efficient at instantiating small Wasm modules that startup times often dropped below one millisecond. In JetStream 2, timing used Date.now() with millisecond precision, meaning sub-1ms times were recorded as zero. The scoring formula, Score = 5000 / Time, then produced an infinite score—a mathematical absurdity that broke comparisons. The team had to patch JetStream 2.2 to cap the score at 5000, but this was only a band-aid. JetStream 3 eliminates the problem entirely with high-precision timing and a new scoring model.
4. High-Precision Timing for Realistic Metrics
To avoid the infinity pitfall, JetStream 3 uses high-resolution timers (e.g., Performance.now()) that capture sub-millisecond differences. This allows the benchmark to distinguish between, say, a 0.3 ms startup and a 0.5 ms startup, which would have both been zero in the old suite. The result is a more granular and fair comparison, especially for engines like JavaScriptCore that aggressively optimized the startup path. It also prevents score inflation and makes every micro-optimization visible.
5. A New Scoring Methodology
JetStream 3 moves away from a simple inverse-time score. Instead, it uses a geometric mean of normalized sub-scores, preventing any single near-zero time from dominating the result. The new approach also accounts for the fact that Wasm startup is no longer a separate concern—it's part of a holistic performance profile. By blending startup and runtime into unified workloads, the benchmark rewards engines that deliver consistent speed across the entire lifecycle of an application.
6. Real-World Workloads at Scale
Earlier benchmarks often used small, synthetic tests that didn't reflect the complexity of modern web apps. JetStream 3 introduces larger, more representative workloads—including interactive visualizations, data processing pipelines, and multi-module Wasm applications. This shift helps engine developers identify bottlenecks that matter to actual users, such as memory management in long-running sessions or rendering latency under heavy JavaScript load.
7. Collaboration Across Browser Vendors
JetStream 3 is the result of a joint effort between the WebKit (Apple), Chrome (Google), and Firefox (Mozilla) teams. This cross-vendor cooperation ensures the benchmark is fair, transparent, and not biased toward any single engine. Each team contributed subtests and reviewed the methodology. The WebKit team, for instance, focused on Wasm instantiation optimizations, while Google's V8 team improved timing precision. The shared codebase means all engines can be tested under identical conditions.
8. How JavaScriptCore Contributed
The WebKit team's JavaScriptCore (JSC) engine underwent significant work to make Wasm instantiation nearly instantaneous. By analyzing the warmup path and leveraging tiered compilation, JSC reduced startup overhead to tens of microseconds—a fraction of what it was in the JetStream 2 era. These improvements are now captured accurately by JetStream 3's high-precision timers. The result is a proof that aggressive engineering on the startup path pays off in real-world page loads where Wasm is used.
9. Implications for Web Developers
For developers, JetStream 3 provides a more reliable yardstick to measure how their applications will perform across browsers. If an application uses Wasm for image decoding or cryptography, JetStream 3's integrated workloads will better reflect the actual user experience. Developers can also use the suite to identify which engine offers the best performance for their specific stack. The benchmark's open-source nature encourages contributions and customization.
10. The Future of Browser Benchmarking
JetStream 3 is not a final destination but a stepping stone. As the web evolves—with new features like WebAssembly GC, thread support, and streaming compilation—benchmarking must keep pace. The collaborative model behind JetStream 3 sets a precedent for future updates, ensuring that benchmarks remain relevant and challenging. The WebKit team hopes this release inspires other engine developers to push performance boundaries while keeping measurements honest and useful.
JetStream 3 represents a thoughtful evolution in performance measurement. By retiring outdated metrics, solving the infinity problem, and embracing real-world complexity, it gives browser engineers and web developers a tool that truly reflects the modern web. Whether you're tuning an engine or optimizing an app, this suite is your new best friend.