基于紧耦合加速器的高性能Java压缩系统  

High-performance Java compression system based on tightly coupled accelerator

在线阅读下载全文

作  者:王雪[1,2] 李文青 张婷婷 张福新 王剑[1,2] 敖琪 WANG Xue;LI Wenqing;ZHANG Tingting;ZHANG Fuxin;WANG Jian;AO Qi(State Key Lab of Processors,Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100190;University of Chinese Academy of Sciences,Beijing 100049;Loongson Technology Corporation Limited,Beijing 100190)

机构地区:[1]处理器芯片全国重点实验室(中国科学院计算技术研究所),北京100190 [2]中国科学院大学,北京100049 [3]龙芯中科技术有限公司,北京100190

出  处:《高技术通讯》2024年第4期386-395,共10页Chinese High Technology Letters

基  金:国家重点研发计划(2022YFB3105103)资助项目。

摘  要:Java无损压缩应用广泛,尽管软件算法在不断改进,但仍然存在压缩速度慢、耗时严重等问题。本文使用领域前沿的紧耦合框架集成压缩加速器的处理器平台,设计了一个高性能Java压缩系统,在Java虚拟机(JVM)内部实现了对紧耦合无损压缩加速器的封装,并为其提供了轻量级运行时环境。本系统可以有效减少通信开销、避免数据拷贝问题,具有易编程、快速压缩的特点,充分发挥加速器给Java压缩带来的性能优势。实验结果表明,此系统大幅提升了Java压缩性能,压缩速度达到主流Java Gzip软件压缩的63倍,最高可达247倍,且在大数据集下性能提升更显著。Java lossless compression is becoming pervasive in a broad range of software systems.Although the compression algorithm has been improved,the low compression speed of Java lossless compression is still a problem.To solve this problem,a high-performance java compression system is proposed.The proposed system adopts the TCADer framework to integrate the CPU with a tightly coupled compression accelerator.Also,the packaging of the compression accelerator is implemented inside the Java virtual machine(JVM).Furthermore,a lightweight runtime environment is also provided for hardware management.In this way,This system can effectively reduce communication overhead and avoid data copy costs,and has the characteristics of easy programming.Experimental results show that this system can greatly improve the performance of Java compression with an average of 63×(up to 247×)speedup over the mainstream Java Gzip compression.Moreover,this system also shows good performance with larger files.

关 键 词:Java压缩 无损压缩 紧耦合加速器 Java虚拟机(JVM) 

分 类 号:TP312.2[自动化与计算机技术—计算机软件与理论] TP311.52[自动化与计算机技术—计算机科学与技术]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象