Evaluating the memory overhead required for COMA architectures
ISCA '94 Proceedings of the 21st annual international symposium on Computer architecture
COMA-F: a non-hierarchical cache only memory architecture
COMA-F: a non-hierarchical cache only memory architecture
The SPLASH-2 programs: characterization and methodological considerations
ISCA '95 Proceedings of the 22nd annual international symposium on Computer architecture
Supercomputing '95 Proceedings of the 1995 ACM/IEEE conference on Supercomputing
Memory consistency and event ordering in scalable shared-memory multiprocessors
ISCA '90 Proceedings of the 17th annual international symposium on Computer Architecture
HPCA '95 Proceedings of the 1st IEEE Symposium on High-Performance Computer Architecture
Bus-based COMA-reducing traffic in shared-bus multiprocessors
HPCA '96 Proceedings of the 2nd IEEE Symposium on High-Performance Computer Architecture
The impact of shared-cache clustering in small-scale shared-memory multiprocessors
HPCA '96 Proceedings of the 2nd IEEE Symposium on High-Performance Computer Architecture
Reducing the Replacement Overhead in Bus-Based COMA Multiprocessors
HPCA '97 Proceedings of the 3rd IEEE Symposium on High-Performance Computer Architecture
Efficient memory simulation in SimICS
SS '95 Proceedings of the 28th Annual Simulation Symposium
Hi-index | 0.01 |
The performance of a COMA multiprocessor greatly depends on the efficiency of the large node caches, the attraction memories. When more than one processor share an attraction memory its behavior is changed.From experiments with program-driven simulation we have found that clustering may improve the performance of the attraction memory significantly. Traffic is reduced, and the miss rates are lower for shared attraction memories. However, clustering may introduce contention for the attraction memory that may ruin any potential performance gain from increased attraction memory hit rate. Provided enough local bandwidth, application execution can remain efficient at higher memory pressure in clustered systems than in systems with single processor nodes. At very high memory pressure some applications change behavior and start suffering from clustering. This is caused by conflict misses due to the relatively lower associativity of the shared attraction memory.