JetStream 3.0 Launches: A 'Fundamental Shift' in Browser Benchmarking for the Modern Web
Breaking: JetStream 3.0 Released
In a joint announcement today, Apple, Google, and Mozilla launched JetStream 3.0, a major overhaul of the cross-browser benchmark suite. The update introduces new WebAssembly tests and addresses a critical 'infinity problem' that had rendered previous benchmarks obsolete for cutting-edge engines.
'JetStream 3 represents both a refresh and a fundamental shift in how we measure performance, particularly regarding WebAssembly and the scale of modern web applications,' a WebKit engineer told reporters. The collaborative effort aims to keep pace with the rapidly evolving web, where benchmarks quickly become outdated.
Background: The Evolution of WebAssembly Benchmarking
When JetStream 2 debuted, WebAssembly was still in its infancy, primarily used for large C/C++ projects compiled from asm.js. The original suite scored WebAssembly in two phases: startup and runtime, anticipating long one-time compilation costs followed by high throughput.
But browser engines—especially WebKit's JavaScriptCore—have become so efficient at instantiating Wasm modules that startup times have plummeted. For smaller workloads, WebKit's startup effectively reached zero seconds, creating a mathematical anomaly in the scoring formula.
The 'Infinity Problem' That Drove the Update
JetStream 2 used the formula Score = 5000 / Time, with time measured via Date.now(). When measured time dropped to 0 ms—as happened for smaller Wasm modules—the score became infinity, breaking the entire benchmark.
'Getting an infinite score sounds like a victory, but it was a clear sign that browser engines had outgrown JetStream 2’s Wasm subtests,' the WebKit engineer explained. The team was forced to patch JetStream 2.2 by capping scores at 5000, a temporary fix that masked the underlying issue.
Modern web applications now integrate WebAssembly into critical paths—libraries, image decoders, and UI frameworks. A microbenchmark showing 'zero' startup time no longer reflects real-world performance, as Wasm often needs to load quickly on every page visit.
What This Means for Browser Performance
JetStream 3.0 completely reworks the Wasm subtests, replacing the outdated startup/runtime split with workloads that better represent contemporary usage patterns. The suite now includes larger, more realistic scenarios where startup time matters but is not the sole metric.
For developers, this means benchmarks will more accurately predict real-world browser performance. 'When the most accessible optimizations have been addressed in a benchmark, subsequent optimizations tend to become less general and more specific to that exact workload,' the engineer noted. 'JetStream 3 directly tackles that problem.'
The update also reflects a broader industry shift: browsers now compete on how quickly they can handle complex, mixed workloads involving JavaScript and WebAssembly together. The infinity problem may be gone, but the race for faster, more responsive web applications is just beginning.