Native Benchmarks vs Web Benchmarks

Benchmarks can be broadly divided into those you install and run on the operating system and those you run directly in the browser. An installed benchmark can be built with code close to the hardware to measure performance near the limit of the device, and it has the advantage of making broad use of operating system features and sensors. On the other hand, you have to download it, install it and grant permissions, and you need to find the right installer separately for each device. A web benchmark runs the same way on a PC, laptop, smartphone or tablet just by opening an address, so it is highly accessible. It is more accurate to see the two approaches as tools with different purposes rather than as a question of which is better. So it matters more to know which question each one answers than to say which is more accurate. Because a web benchmark needs no separate install and makes it easy to match conditions across devices, it is also convenient for checking the devices of a family or a team all at once.

The biggest difference is what is being measured. An installed benchmark generally shows the potential of the hardware itself, while a web benchmark shows the performance after passing through the browser engine and its security policies. Because so much work has moved to web apps these days, performance in the browser is itself the real user experience. But even on the same hardware, values change with the browser type and version, JavaScript optimization, Worker scheduling and timer precision. Therefore the scores of the two approaches have different units and standards, and the numbers should not be converted or compared directly. Neither one is right or wrong; they are values that answer different questions. It is natural to see the detailed items an installed tool provides and the broad accessibility of the web as complementing each other.

So how should you use them? If your purpose is choosing hardware or tuning to the limit, an installed benchmark fits, and if you want to quickly check whether your current browser environment feels smooth, or measure on a device where installing is difficult, a web benchmark is convenient. In Web Bench, the quick test takes about 30 seconds and the standard test about 2–3 minutes, and results are sent to a server only with consent to join the rankings. You can also compare your last 20 results to see the change before and after altering settings or switching browsers. Whichever tool you use, the most important rule is to compare the same tool, the same version and the same conditions, and to accept every score as a reference value. Even if you use several tools together, do not mix their scores, and building the habit of reading what each tool tells you separately will reduce misunderstandings. The result card can be saved as a PNG, which also makes it easy to share with others.