Hybrid coherence for scalable multicore architectures

Accelerator architecture;hybrid architectures;memory model;cache coherence

Kelm, John H.

关键词: Accelerator architecture;    hybrid architectures;    memory model;    cache coherence;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/18272/Kelm_John.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

This work describes a cache architecture and memory model for 1000+ core microprocessors.Our approach exploits workload characteristics and programming model assumptions to build a hybrid memory model that incorporates features from both software-managed coherence schemes and hardware cache coherence.The goal is to achieve the scalability found in compute accelerators, which support relaxed ordering of memory operations and programmer-managed coherence, while providing a programming interface that is akin to the strongly ordered cache coherent memory models found in general-purpose multicore processors today.The research presented in this dissertation supports the following thesis: To be scalable and programmable, future multicore systems require a cached, single-address space memory hierarchy.A hybrid software/hardware approach to coherence management is required to support such a memory hierarchy in 1000+ core processors and is achievable only by leveraging the characteristics of target applications and system software.We motivate a hybrid memory model and present our approach to addressing the challenges facing such a model.We discuss and evaluate a scalable 1024-core architecture, workloads that we see as targets for such an architecture, a memory model that relies on software management of coherence, and scalable hardware coherence schemes. Using these components, we develop the software and hardware support for a hybrid memory model.We demonstrate that our techniques can be used to reduce hardware design complexity, to increase software scalability, or to combine the two.