Multilayer Adaptations And Optimizations Of Android 2D Graphics Library For The Loongson Architecture

Android operating system has occupied 82% market share due to its rich software functions and significant user experience. Because most of the mobile application processors are developed upon the ARM architecture, the native Android makes optimization just for ARM CPUs, which promotes the user experience greatly. On contrast, Google has only released unoptimized Android implementations for other architectures..Based on detailed analysis of Skia Graphics library, the thesis finds that it takes most of the time for loogson 2H processor to complete the filling of the stored data and the implementation of the core algorithms with Perf which is the System bottleneck analysis tool. Firstly, considering the 64 width advantage of loogson multimedia registers, the thesis modifies the way that 2H used to fill the stored data and improves the filling efficiency. Secondly the thesis accelerates the core algorithm with the feature of Loogson Multimedia Instrcution set, which makes the 2H platform deal with the parallel computing of multiple pixels and optimizes the execution efficiency of Graphic image processing. At last, the thesis uses the cache prefetch technology to reduce the memory access latency and increase efficiency of memory access.The testing process is under default configuration of loogson 2H. After the optimization the canvas filling performance has been improved by 25% and the image rendering performance has been improved by 29.8% according to Oxbenchmark. While according to results from Skbenchmark, the Graphics rendering has been promoted by 20% and the Image manipulation performance has been promoted by 40%.