Practical Lessons from Client Performance Optimization in a Mobile Game Project

The author reflects on client performance testing in a mobile basketball game, highlighting personal experiences and valuable lessons that can be transferred across teams, without delving into basic theory or specific tool tutorials.

01 Recognizing the Long‑Term Nature of Performance Optimization – Early attempts to set strict production standards failed, as performance knowledge evolves and project goals (e.g., targeting low‑end Philippine devices) shift, making performance standards fluid and optimization a continuous effort.

02 Adding Data Points for Major Performance Decisions – Logs and lightweight data points differ: logs are verbose and rarely collected post‑release, while targeted data points capture essential device and performance information, enabling informed decisions such as choosing OpenGLES 3.0 and ETC2 compression for low‑end devices.

03 Using Switches for Performance Features – Feature toggles allow developers to enable or disable specific logic during profiling, facilitate rapid testing on low‑end hardware, and provide a safety net for unverified optimizations in production builds.

04 Selecting Appropriate Test Packages – Different build configurations (development build, performance‑test build, performance‑test‑with‑log build) affect frame rates and data accuracy; using non‑development builds yields more realistic performance metrics.

05 Making Test Behavior Close to Real Users – Continuous multi‑round testing, realistic session lengths, and monitoring performance across several matches reveal issues (e.g., memory growth, unexpected function calls) that single‑run tests miss.

06 Cautiously Evaluating Costly Optimizations – Trade‑offs such as pre‑loading assets improve frame stability but increase memory usage and load time; careful measurement and selective application prevent regressions, as illustrated by optimizing ball‑trajectory loading.

07 Synchronizing Optimization Information – Clear communication of performance changes (e.g., disabling background rendering for full‑screen UI) prevents bugs like black screens and reduces the time spent troubleshooting by the wider team.

08 Conclusion – The article summarizes scattered experiences, emphasizing that sharing practical testing and optimization insights can help other teams avoid similar pitfalls.