检索规则说明:AND代表“并且”;OR代表“或者”;NOT代表“不包含”;(注意必须大写,运算符两边需空一格)
检 索 范 例 :范例一: (K=图书馆学 OR K=情报学) AND A=范并思 范例二:J=计算机应用与软件 AND (U=C++ OR U=Basic) NOT M=Visual
作 者:石强[1] 赵荣彩[1] 高亚虎 王军[1] SHI Qiang;ZHAO Rongcai;GAO Yahu;WANG Jun(Information Engineering University,Zhengzhou 450001,China)
机构地区:[1]信息工程大学
出 处:《信息工程大学学报》2019年第3期308-312,共5页Journal of Information Engineering University
基 金:国家自然科学基金资助项目(61472447)
摘 要:SIMD指令在大多数现代处理器架构中是必需的拓展部件。使用SIMD指令能同时对一组数据执行相同操作,从而实现空间上的并行性,显著提高处理性能。因此SIMD指令在二进制文件中指令的比例显著提高。但大部分动态二进制翻译器忽视了SIMD指令的翻译,源平台的SIMD指令常被翻译为一系列的标量指令,即要求目标平台拥有对应的SIMD指令支持此类并行计算。文章以申威国产处理器为实现平台,提出基于QEMU系统翻译SIMD指令优化方法,包括两个方面,一是借助修改已在QEMU中实现的helper函数,二是使用新引进的向量中间表示提高翻译系统QEMU对SIMD指令的翻译效率。前端为IA32平台,优化后的QEMU翻译对源平台的文件取得较好的加速效果。As a necessary extension in most modern processor architectures,SIMD instructions can perform the same operations on a set of data at the same time to achieve spatial parallelism and significantly improve processing performance.As a result,the SIMD instruction has a significantly higher percentage of instructions in the binary file.However,most of the dynamic binary translators ignore the translation of SIMD instructions,such as the more popular QEMU translation system.The source platform SIMD instruction is often translated into a series of scalar directives,even if the target platform has the corresponding SIMD instructions to support the parallel computing.In this paper,domestic processors are used for the realization of SW platform,and we propose a SIMD instruction optimization method based on QEMU system translation.It consists of two aspects:first,the helper function which has been implemented in QEMU is modified;second,the newly introduced vector intermediate representation is used to improve the efficiency of SIMD instruction translation of QEMU.The front-end is IA32 platform,and the optimized QEMU has better speedup effect on the source files.
分 类 号:TP314[自动化与计算机技术—计算机软件与理论]
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在链接到云南高校图书馆文献保障联盟下载...
云南高校图书馆联盟文献共享服务平台 版权所有©
您的IP:216.73.216.7