Characterization of Multi-core Processors for Power-Estimation at System-level

用于系统级功耗估计的多核处理器的表征

• 批准号: 166220900
• 负责人: Professor Dr.-Ing. Wolfgang Nebel
• 金额: --
• 依托单位: Department of Computer Science and Engineering
• 依托单位国家: 德国
• 项目类别: Research Grants
• 财政年份: 2010
• 资助国家: 德国
• 起止时间: 2009-12-31 至 2013-12-31
• 项目状态: 已结题
• 来源: https://gepris.dfg.de/gepris/projekt/166220900?language=en
• 关键词: Characterization; Multi; core; Processors; Power

The overall goal of this cooperative project is to advance the understanding of the relation between power consumption of multi-core processors, their architecture und the applications running on them. This will lead to compact yet accurate models of their power dissipation under consideration of architectural parameters and under consideration of the particular application software. These models can be used to fast yet sufficiently accurate estimate the power consumption of different system configurations and hence enable an efficient design space exploration of power aware embedded systems. Finally the project will contribute to the competitiveness of the European semiconductor, system integrator and EDA industry as well as the Indian design industry by addressing the problem of platform-based design space exploration (DSE) under consideration of power and performance constraints early in the design process. High performance usually causes high power consumption. As a consequence, a main challenge in today s embedded system design is to find the perfect balance between performance and power, which is addressed by this proposal. From a scientific perspective this project s objective for achieving the above goal is to develop a characterization-methodology with focus on embedded multi-core processor power dissipation on a very high level of abstraction, allowing fast and early power estimations. The basic block granularity at TLM2 proposed in this project serves as best trade-off between accuracy and estimation time. Eventually, the project will help to establish a closer working cooperation between the two project partners University of Oldenburg and IIT Delhi, strengthening their cooperation through a joint R&D project, and at the same time increasing mobility of the involved researchers.

该合作项目的总体目标是促进对多核处理器的功耗、其架构以及在其上运行的应用程序之间的关系的理解。这将导致在考虑架构参数和特定应用软件的情况下，其功耗的紧凑而准确的模型。这些模型可以用来快速但足够准确地估计不同的系统配置的功耗，从而使一个有效的设计空间探索的功耗感知嵌入式系统。最后，该项目将有助于欧洲半导体，系统集成商和EDA行业以及印度设计行业的竞争力，通过解决基于平台的设计空间探索（DSE）的问题，在设计过程的早期考虑功率和性能限制。高性能通常会导致高功耗。因此，在当今的嵌入式系统设计的一个主要挑战是找到性能和功耗之间的完美平衡，这是解决这个建议。从科学的角度来看，该项目的目标是实现上述目标，开发一种表征方法，专注于嵌入式多核处理器的功耗在一个非常高的抽象层次，允许快速和早期的功耗估计。本项目中提出的TLM2的基本块粒度是准确性和估计时间之间的最佳折衷。最终，该项目将有助于在奥尔登堡大学和德里理工学院这两个项目合作伙伴之间建立更密切的工作合作，通过联合研发项目加强合作，同时增加参与研究人员的流动性。

项目成果

Impact of Data Sharing on Co-Running Embedded Applications in Multi-core System

数据共享对多核系统中并行运行嵌入式应用的影响

• DOI: 10.1109/pdp.2015.88
• 发表时间: 2014
• 期刊: 2015 23rd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing
• 作者: A. Korotaeva;W. Nebel;M. Balakrishnan
• 通讯作者: M. Balakrishnan