Collaborative Research: CSR-PSCE, TM: Effective Resource Sharing and Coordination inside Multicore Processors for High Throughput Computing

多核处理器资源管理面临的主要挑战之一是现有的操作系统（无论是传统的单核操作系统还是SMP操作系统）无法有效地处理多核处理器中的新复杂性。为了应对这一挑战，合作者将开展三个密切相关的项目。(1)基于操作系统的高速缓存划分的混合系统设计和实现将以最小的硬件复杂度提供对共享高速缓存的有效软件管理，并将很好地定义共享高速缓存管理器的硬件/软件接口。(2)合作者将在操作系统内核中设计和实现调度算法，以有效地分配CPU，缓存和内存带宽资源，以多核处理器中的多编程作业。(3)一个数据对象局部性感知的缓存分区设计和实现将区分对象的局部性优势，并作出有效的缓存分配决策。本项目的智力挑战有三个方面：（1）混合系统设计涉及硬件和下划线操作系统之间的复杂交互，并要求对现有的系统结构和创新的深刻理解，以提高体系结构和操作系统内核。(2)多核处理器中基于操作系统的调度是系统研究中的一个基本而复杂的问题。(3)调度和对象协调算法的系统实现需要很多创造性的想法，以便在内核中无缝集成。该项目的广泛影响预计将是重大的。在许多应用领域，及时地需要解决多核处理器中的关键问题以显著提高性能。对本科生和研究生的研究培训将解决IT和计算机行业缺乏强有力的系统专业人员的问题。

One of the main challenges in multi-core processor resource management is thatexisting operating systems (either conventional OS for single core or OS for SMP) arenot able to effectively handle the new complexities in multi-core processors.In order to address this challenge, the collaborators will conduct three closely relatedprojects. (1) A hybrid system design and implementation for OS-based cachepartitioning will provide efficient software management of shared caches with minimumhardware complexity, and will well define hardware/software interface of shared cachemanagement. (2) The collaborators will design and implement scheduling algorithms inOS kernels to effectively allocate CPU, caches and memory bandwidth resources tomultiprogramming jobs in multi-core processors. (3) A data object locality-aware cachepartitioning design and implementation will distinguish the locality strengths of objectsand make effective cache allocation decisions.The intellectual challenges of this project are threefold: (1) Hybrid system designinvolves complex interactions between hardware and the underlined operating system,and demands insightful understanding of existing system structures and innovation toenhance both architecture and the OS kernels. (2) OS-based scheduling in multi-coreprocessors is a fundamental and complex problem in system research. (3) Systemimplementation of proposed algorithms for scheduling and object coordination demandsa lot of creative ideas for their seamless integration in the kernels. The broader impactof this project is expected to be significant. Solutions to address critical issues forsignificant performance improvement in multi-core processors are timely demanded inmany application areas. The research training to both undergraduate and graduatestudents will address the concerns of lacking strong system professionals in IT andcomputer industries.