Kotlin Multiplatform vs Native Development: A Performance Analysis for Android and iOS

Sammanfattning

This thesis investigates the performance differences between Kotlin Multiplatform (KMP) and native implementations across the mobile platforms Android and iOS. KMP is a cross-platform framework designed to enable code sharing across different platforms, including mobile, web, and desktop applications. The objective is to identify scenarios where KMP outperforms or underperforms compared to native implementations, analyze factors contributing to performance variations, and determine the best use cases for the adoption of KMP as a framework. The study uses a controlled experimental approach, testing both KMP and native implementations across ten different devices (five Android and five iOS devices) to ensure consistency in the results. The findings reveal substantial platform-specific variations in performance. KMP displays superior performance in image rendering tests across both platforms (58% to ∼83% faster) and shows noteworthy advantages on iOS for startup times (∼57% faster) and CPU (Central Processing Unit) performance (∼81% faster). However, KMP exhibits slower performance on Android for startup times (∼44% slower) and CPU operations (∼26% slower). The study also reveals substantial differences in binary application size. While KMP shows no particular differences on Android, the iOS implementations ended up with ∼33 MB and ∼1 MB, respectively, for KMP and native implementations. The results suggest that KMP is particularly well-suited for applications requiring a consistent user interface, those with significant image rendering requirements, and projects prioritizing development efficiency through code sharing. However, the choice to adopt KMP should be made with careful consideration of platform-specific requirements and performance priorities.