GPU异构系统中的存储层次和负载均衡策略研究
GPU이구계통중적존저층차화부재균형책략연구
Research on Memory Hierarchy and Load Balance Strategy in Heterogeneous System Based on GPU
  • 万方数据
    国防科技大学学报 국방과기대학학보
    JOURNAL OF NATIONAL UNIVERSITY OF DEFENSE TECHNOLOGY
    2009年 5期 38-43 ,共6页

    马安国%成玉%唐遇星%邢座程

    마안국%성옥%당우성%형좌정

    GPGPU%存储层次%负载均衡策略%流计算%任务划分 GPGPU%存儲層次%負載均衡策略%流計算%任務劃分
    GPGPU%존저층차%부재균형책략%류계산%임무화분
    GPGPU%memory hierarchy%load balance strategy%stream computing%task division
    GPU体系结构的革新和相应开发平台的发展使得GPU广泛地应用于科学计算领域.通过深入地分析GPU体系结构和存储层次的优缺点以及GPU上的关键性能特征,阐明了GPU体系结构、编程模型和存储层次之间的关系.针对GPU异构系统上的应用映射提出三种基本负载均衡优化策略:预取、流化、任务划分.试验结果揭示了不同的优化因子与优化效率之间的具体关联.
    GPU체계결구적혁신화상응개발평태적발전사득GPU엄범지응용우과학계산영역.통과심입지분석GPU체계결구화존저층차적우결점이급GPU상적관건성능특정,천명료GPU체계결구、편정모형화존저층차지간적관계.침대GPU이구계통상적응용영사제출삼충기본부재균형우화책략:예취、류화、임무화분.시험결과게시료불동적우화인자여우화효솔지간적구체관련.
    Owing to the revolution of GPU architecture and improvement of developing platforms, GPU is widely used in scientific computing nowadays. Relationships truing GPU architecture, programming model and memory hierarchy are illustrated by analyzing memory hierarchy and exploring key performance features of GPU. Three basic load balance strategies on mapping applicatiom onto GPU are presented: Prefetch, stream computing, task division. The effective rehtionships among different factors and optimization efficiency are tested and exposed by experiments.
    원문보기 원문다운로드 사이트로이동
저자의 최근 논문