by Martin Brinkmann on March 28, 2019 in Internet – No comments
WebKit launched JetStream 2, a new benchmark suite to measure the performance of JavaScript and WebAssembly operations in web browsers recently.
Browser benchmarks were a huge thing back when Google Chrome launched. Google used them to demonstrate how much faster Chrome’s JavaScript performance was when compared to Internet Explorer or Firefox, the two main browsers back in the day.
One of the effects was that browser makers started to optimize and improve the performance of JavaScript; this resulted in benchmarks becoming less important as speed improved in browsers.
JetStream 2
JetStream 2 tests JavaScript and WebAssembly performance of web browsers as well as the performance of other functions such as Web Workers. Just hit the start button on the JetStream 2 benchmark website to test the browser.
JetStream 2 also includes a new set of benchmarks that measure the performance of Web Assembly, Web Workers, Promises, async iteration, unicode regular expressions, and JavaScript parsing.
The browser benchmark runs 64 tests, some of which come from other benchmarks such as JetStream, SunSpider, or Octane. The development team describes each of the tests that JetStream 2 runs on this page.
The team ran the benchmark on a MacBook Pro to compare the results of Safari, Chrome and Firefox. Safari took the crown in the test followed by Chrome (about 8% slower) and Firefox (about 68% slower).
I decided to ran the test on a Windows machine. While I could not run Safari on Windows, I ran the benchmark in recent stable versions of Chrome, Firefox and Microsoft Edge.
Chrome managed to get a score of about 105, Firefox a score of 78 in the benchmark. Firefox Nightly got an Infinity score instead which suggests that something broke while the test was running. The test did not complete in Microsoft Edge and I had to stop it as it would not even finish the first test of the benchmark suite.
Closing Words
The new benchmark suggests that Mozilla has some work to do to close the performance gap according to the benchmark. Benchmarks don’t necessarily relate to real-world performance though.
Now You: Did you run the benchmark? How did browsers installed on your device perform?