面向GPU的异构流计算系统资源管理与调度机制  

Resource management and scheduling mechanism for heterogeneous stream computing system oriented to GPU

作　　者：顾平莉 李融 李常宝[1] 刘忠麟[1] GU Ping-li;LI Rong;LI Chang-bao;LIU Zhong-lin(North China Institute of Computing Technology,Beijing 100083,China)

机构地区：[1]华北计算技术研究所

出　　处：《信息技术》2019年第12期15-19,共5页Information Technology

基　　金：国家科技重大专项核心电子器件、高端通用芯片及基础软件产品(核高基)(2015ZX01040201)

摘　　要：文中针对当前广泛使用的大数据流计算系统Storm无法对GPU资源进行管理、调度的问题,提出一种包含GPU在内的流计算系统资源状态管理与调度机制,实现分布式集群环境下的GPU资源状况获取、GPU负载度检测、包含GPU资源的系统资源调度,形成分布式集群环境下对通用计算资源与GPU整体的负载状态评估与按需调度能力。Aiming at the problem that the currently widely used big data stream computing system cannot manage and schedule GPU resources,a resource state management and scheduling mecha-nism of the stream computing system including the GPU is proposed to realize the GPU resource status acquisition,GPU load degree detection,and system resource scheduling including GPU resources in a distributed cluster environment,and to form a load state assessment and on-demand scheduling capability for general computing resources and GPUs in a distributed cluster environment.

关 键 词：流计算 异构计算 GPU STORM 资源调度 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]