计算机工程与科学2018，Vol.40Issue(4)：602-608,7.DOI:10.3969/j.issn.1007-130X.2018.04.005
一种基于频度统计的动态二进制翻译优化方法
A dynamic binary translation optimization method based on frequency statistics
李男 1庞建民 1单征1
作者信息
- 1. 解放军信息工程大学,河南郑州450002
- 折叠
摘要
Abstract
In a dynamic binary translation process,putting the code with high executing frequency into the translating cache for long time and extending the amount of code executed by the translator at one time is an effective way to reduce the context switching overhead and improve the system efficiency.Therefore,we propose the optimization clue of "identifying the hot code→constructing the super block →improving the T-Cache management method",and design a hot code identification algorithm based on frequency statistics,which regard the basic block whose executing frequency is more than the preset threshold value as well as its subsequent basic blocks as the hot codes.Besides,we propose an idea of constructing super blocks generated by physically chaining the basic blocks in the hot code and hence owns a larger capacity in order to support the T-Cache system.Based on these,the original search method and replacement strategy of the T-Cache system are improved.Experimental results verify the correctness and the performance of the optimization method.On the domestic ShenWei processor platform,the method makes the standard test set SPEC 2006 achieve an average performance improvement of 9.34 ％.关键词
二进制翻译/热代码/超块/T-Cache/QEMUKey words
binary translation/hot code/super block/T-Cache/QEMU分类
信息技术与安全科学引用本文复制引用
李男,庞建民,单征..一种基于频度统计的动态二进制翻译优化方法[J].计算机工程与科学,2018,40(4):602-608,7.
