CSR: Small: Collaborative Research: Transparent and Energy-Efficient Speculation on NUMA Architectures for Embedded Multiprocessor Systems

高端嵌入式系统正在转向多核体系结构，以满足其应用程序所需的性能和性能/瓦特要求。此外，随着嵌入式系统对更多计算密集型功能的需求增加，这些多核体系结构将演变为多核系统，以提高性能或性能/面积/瓦。这些系统通常被组织为基于集群的非一致内存访问(NUMA)体系结构，为程序员提供共享内存抽象。也就是说，简单核心被分组为共享本地互连和内存的集群，然后使用可扩展的片上网络介质来复制和互连这些集群。该项目研究了这些新兴的NUMA架构给嵌入式系统带来的主要挑战之一：提供高效、节能和方便的同步和通信机制。特别是，它提出了新的解决方案，基于对投机性同步的硬件支持和使这种投机性透明的软件支持。衡量解决方案有效性的重要指标包括吞吐量、易用性、系统能耗和架构简单性。嵌入式系统在包括智能手机、汽车系统、安全和其他环境智能系统在内的广泛应用中变得无处不在。不断增长的计算需求导致了更复杂的产品，因此在满足严格的设计限制方面面临着更大的挑战，特别是吞吐量/瓦特。这些系统通信和同步数据的方式的改进可以在改进的功能、实用性和耐用性方面产生重大影响。该项目是一项国际合作，结合了PI在片上网络架构、嵌入式系统设计和存储器同步方面的专业知识。该提案的更广泛影响包括新课程的开发，面向研究生和本科生女性以及代表性不足的少数群体，以及国际机构之间的学生交流。

High-end embedded systems are turning to multicore architectures to meet the performance, and performance/Watt, demands required for their applications. Moreover, as the demand for more compute-intensive capabilities for embedded systems increases, these multicore architectures will evolve into many-core systems for improved performance or performance/area/Watt. These systems are often organized as cluster-based Non-Uniform Memory Access (NUMA) architectures that provide the programmer with a shared-memory abstraction. That is, simple cores are grouped into clusters sharing local interconnection and memory and these clusters are then replicated and interconnected using a scalable network-on-chip medium. This project investigates one of the principal challenges presented by these emerging NUMA architectures for embedded systems: providing efficient, energy-effective, and convenient mechanisms for synchronization and communication. In particular, it proposes new solutions based on hardware support for speculative synchronization, and software support to make such speculation transparent. Important metrics for measuring the effectiveness of the solutions include throughput, ease of use, system energy consumption, and architectural simplicity. Embedded systems are becoming ubiquitous over a broad range of applications, including smart phones, automotive systems, security, and other ambient intelligence systems. Increasing computational demands have led to more sophisticated products and therefore increased challenges in meeting tight design constraints, particularly throughput/Watt. Improvements to the way these systems communicate and synchronize data can have a substantial impact in terms of improved functionality, utility, and durability. The project is an international collaboration that combines PI expertise in Network-on-Chip architectures, embedded system design, and memory synchronization. Broader impacts of the proposal include new course development, outreach to graduate and undergraduate women and under-represented minorities, and student exchanges between international institutions.