Smartphone Performance and Heat Test

A smartphone runs its CPU and GPU inside a small, thin body, so it is hard to cool quickly. That is why it is common for a phone to be fast at first and then slow down after extended use. Stuttering when you play a game for a long time or record video can have the same cause. Web Bench runs as it is in a smartphone browser and needs no install. A quick test of about 30 seconds shows the current speed, and a standard test of about 2–3 minutes looks broadly at CPU, memory, browser, storage and the supported GPU tests. Keep in mind that peak speed for a moment is different from the speed you can sustain, and read the results with that in mind. A smartphone benchmark can also be used for a light check on a store demo unit or a friend's phone before you buy. Phones also tend to have strong power-saving policies in the operating system, so results can differ even when you compare with the same apps left open.

Use the stress test to check heat. It runs CPU multi, GPU or both loads for 5 or 10 minutes, records the score every 15 seconds and shows it as a graph. Stability is the lowest recorded score divided by the highest, and the point where the score drops sharply several times in a row is estimated as the start of throttling. This is not a direct reading of a temperature sensor but a value inferred from the change in performance. Background tasks or power-saving settings can create a similar drop, so do not conclude from the curve alone that the device is defective. If the device feels too hot, stop it immediately with the stop button. Phone calls and notifications arriving during a test can disturb the result, so turning on Do Not Disturb also helps.

Before testing, decide whether to unplug or plug in the charger, and repeat under the same conditions. While charging, the heat from the battery and the body increases and results can change. Taking off the case and testing in a cool place reduces the effect of heat, and the screen must stay on with the test tab in front. If you hide the tab or the screen turns off, the segment in progress is discarded and the test pauses. Comparing results in low power mode and normal mode gives you a sense of how much a battery-saving setting reduces performance. Results are sent to a server only with consent to join the rankings and can be downloaded as a PNG card, but check the device model information on the card before sharing it. Even if a result is lower than expected, check your settings and environment first rather than suspecting a defect right away, and measure again under matching conditions. The standard test finishes in about 2–3 minutes, so if heat is a concern, start with the quick test first.