融合CUDA与OPI制导调优的语言在线翻译器自动化并行优化研究  

Research on Automated Parallel Optimization of Online Language Translators Based on CUDA and OPI Guided Optimization

作　　者：刘鹏娟[1] LIU Pengjuan(Xi’an Siyuan University,Xi’an 710038,China)

机构地区：[1]西安思源学院,西安710038

出　　处：《自动化与仪器仪表》2024年第9期230-233,238,共5页Automation & Instrumentation

基　　金：陕西体育局项目《“全民健身”视域下西安国际马拉松赛事发展研究》(2018018)。

摘　　要：针对语言翻译器的并行优化主要依赖于手动调整和优化较为费时费力这一缺点,研究提出一种融合统一计算设备架构(Compute Unified Device Architecture, CUDA)和最佳性能指标(Optimal Performance Index, OPI)制导调优的语言在线翻译器自动化并行优化方法。此次研究选择扩展ParaC编译器,使其能生成CUDA程序,并引入了OPI制导调优,以期在并行处理和性能优化方面达到更好的效果。研究进行了仿真实验,结果显示,传统方法的性能相对于所提方法略低,平均性能达到所提方案的79.98%,最优时达到93.97%。研究为在线翻译器的进一步发展奠定了重要的理论基础,并为其实践应用提供了有力的指导。In response to the drawbacks of manual adjustment and time-consuming optimization in parallel optimization of language translators,this study proposes an automated parallel optimization method for language online translators that integrates CUDA and OPI guided tuning.This study chose to extend the ParaC compiler to generate CUDA programs,and introduced OPI guided tuning to achieve better results in parallel processing and performance optimization.This study conducted simulation experiments and the results showed that the performance of traditional methods was slightly lower than that of the proposed method,with an average performance of 79.98%of the proposed scheme and an optimal performance of 93.97%.This study has laid an important theoretical foundation for the further development of online translators and provided strong guidance for their practical applications.

关 键 词：CUDA OPI制导调优 ParaC语言编译器 并行优化 仿真实验 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]