Mobile Smartphone Speed Test

When a new smartphone is released, there are multiple technology magazines which do reviews of the device. They usually test new features, the cameras, various other sensors and measure the performance by running different synthetic benchmarks. These benchmarks mainly focus on the maximum performance the System-on-a-Chip can deliver in the tested device. Most of the time the user does not use this maximum performance because usual every-day tasks such as sending messages, taking photos or browsing the Internet do not need high computation power. This real-world performance is rarely tested and so there are no numbers to easily compare the real-world speed of smartphones. This thesis presents a two-device solution for measuring the real-world performance of an Android device. The solution consists of two apps: SpeedtestReplay and SpeedtestControl. The tested device runs SpeedtestReplay, the other device acts as measuring tool running SpeedtestControl, which records the other device with the camera. The taken images are analysed to get the time the tested device needs to start the specified app. The measurement is based on the assumption that as soon the tested app is started completely the screen remains static. These apps are used to compare the real-world performance of three devices: Sony Xperia V, Sony Xperia Z1 Compact and Sony Xperia Z3 Compact, three similar devices from different generations. The results show that as expected the two newer devices perform much better but the difference between the Z1 Compact and Z3 Compact is bigger than one might expect from the synthetic benchmark results. An additional test is performed between two devices with identical hardware. One of the devices is freshly reset, the other is used daily. Comparing just the app start times, the difference between the two devices was negligible. When switching between multiple apps there was a bigger difference.