Allen Chen
ABSTRACT
One of the most important concepts that must be understood by students of parallel computer architecture is cache coherence. Processors may change the value of shared variables by writing them in their private caches. These changes must be propagated to the far reaches of the system so that writes to each location are seen in the same sequence throughout the system. Of course, this can be achieved by various protocols, depending on the size of the system and the workload. Students can gain an in-depth understanding of these protocols by simulating them. Our approach is to provide students with an abstract simulator, which can be specialized through inheritance in C++ to model many different protocols. The protocols can be divided into bus based and directory based. Either of these can be invalidation or update. Hybrid and adaptive protocols can also be simulated. Among the protocols covered are MSI, MESI, MOESI, Firefly, Dragon, and a simplified SCI protocol. It would be easy to add additional protocols by subclassing appropriate classes.
- Solihin, Yan, Fundamentals of Parallel Computer ArchitectureGoogle Scholar
- Culler, D. E., Singh, J. P., and Gupta, A. Parallel Computer Architecture. A hardware/software approach. Morgan Kaufmann Publishers, Inc., 1999 Google ScholarDigital Library
- Patterson, D. A., and Hennessy, J. L. Computer Organization and Design: The Hardware/Software Interface. Morgan Kaufman Publishers, Inc., 2004. Google ScholarDigital Library
- Mohit Gambhir, Edward F. Gehringer, and Yan Solihin, "Animations of important concepts in parallel computer architecture," Workshop on Computer Architecture Education, International Symposium on Computer Architecture, San Diego, June 9, 2007. Google ScholarDigital Library
- Cache Coherence Simulator Student's VersionGoogle Scholar
Recommendations
An Interactive MESI Cache Coherence Simulator for Educational Purposes
PCI '16: Proceedings of the 20th Pan-Hellenic Conference on InformaticsIn multi-processor systems data can reside in multiple levels of cache, as well as in main memory. The problem of keeping the data consistent among all caches and memory is known as the cache coherence problem. There are different protocols to solve ...
Boosting performance of directory-based cache coherence protocols with coherence bypass at subpage granularity and a novel on-chip page table
CF '16: Proceedings of the ACM International Conference on Computing FrontiersChip multiprocessors (CMPs) require effective cache coherence protocols as well as fast virtual-to-physical address translation mechanisms for high performance. Directory-based cache coherence protocols are the state-of-the-art approaches in many-core ...
Increasing the effectiveness of directory caches by deactivating coherence for private memory blocks
ISCA '11: Proceedings of the 38th annual international symposium on Computer architectureTo meet the demand for more powerful high-performance shared-memory servers, multiprocessor systems must incorporate efficient and scalable cache coherence protocols, such as those based on directory caches. However, the limited directory cache size of ...
Comments