System-Level Performance Analysis and Optimization for Cross-Platform Execution Environments

Abstract

Cross-platform execution technologies enable legacy applications to run on modern hardware platforms, but their practical performance is often affected by hidden system-level overhead. In real deployments, factors such as memory translation, exception handling, and cache interference play a critical role in overall efficiency. An experimental environment consisting of heterogeneous servers and full-system virtual machines was constructed to analyze runtime behavior under mixed workloads. Profiling results indicate that translation cache locality and synchronization frequency strongly influence execution stability. Based on these observations, several lightweight optimization techniques were introduced at the translation and memory management layers. Evaluation using standard benchmark suites shows that the optimized configuration achieves more stable throughput under long-running workloads. The study suggests that sustainable cross-platform performance depends more on system coordination than on isolated translation speed.