动态二进制翻译中数据预取优化研究*

动态优化是动态二进制翻译研究中一个十分重要的课题，数据预取优化能提高现代处理器体系结构应用程序性能。基于超级块(Superblock)的动态数据预取优化采用软件插桩方式收集应用程序的load访存延迟信息并构造Superblock；然后根据延迟信息以及Superblock数据流分析得出的寄存器定值引用关系，对延迟load指令进行预取优化。通过在龙芯DigitalBridge动态二进制翻译系统上实验验证，数据预取优化可以提高翻译后SPEC2000浮点测试程序代码的平均性能3.3%，开销远小于0.5%。

Data prefetch optimization in dynamic binary translation

Dynamic optimization is an important research subject in binary translation systems. As a way of memory optimization, data prefetching improves application's performance on modern processor architectures. This paper proposed a Superblock based dynamic data prefetching scheme for dynamic translation, it used software instrumentation to collect program's load delay information and built Superblocks. Then performed the data flow analysis on Superblocks to generate the RDUG(register define and use graph). After that, inserted data prefetching instructions for the delinquent loads. Experiments on the Loongson DigitalBridge dynamic binary translation system show that the prefetching scheme can achieve 3.3% improvement on average for SPEC2000 float programs while the overhead is less than 0.5%.