The connection machine
Optimal partitioning of randomly generated distributed programs
IEEE Transactions on Software Engineering
IEEE Spectrum
Stencils and problem partitionings: their influence on the performance of multiple processor systems
IEEE Transactions on Computers
Performance analysis of the connection machine
SIGMETRICS '90 Proceedings of the 1990 ACM SIGMETRICS conference on Measurement and modeling of computer systems
Seismic modeling at 14 gigaflops on the connection machine
Proceedings of the 1991 ACM/IEEE conference on Supercomputing
Approximate performance models of polling systems using Stochastic Petri nets
IEEE INFOCOM '92 Proceedings of the eleventh annual joint conference of the IEEE computer and communications societies on One world through communications (Vol. 3)
Probability and Statistics with Reliability, Queuing and Computer Science Applications
Probability and Statistics with Reliability, Queuing and Computer Science Applications
A Perspective on Iterative Methods for the Approximate Analysis of Closed Queueing Networks
Proceedings of the International Workshop on Computer Performance and Reliability
Fixed-Point Approximations for Distributed Systems
Proceedings of the International Workshop on Computer Performance and Reliability
Fixed Point Iteration in Availability Modeling
Proceedings of the 5th International GI/ITG/GMA Conference on Fault-Tolerant Computing Systems, Tests, Diagnosis, Fault Treatment
Architecture of a massively parallel processor
ISCA '80 Proceedings of the 7th annual symposium on Computer Architecture
| Hi-index | 0.00 |
The performance of the interprocessor communication architecture of the CM-2 isanalyzed. A discrete-time Markov chain model of its network architecture is developed tocompute the message delay introduced by the network architecture. Due to thesynchronous time-division multiplexing nature of the network operation, it is amenable toa discrete-time Markov chain modeling. The analysis yields formulas for response time and several other related performance measures, showing how the performance of thenetwork degrades with the message arrival rate and other parameters. Since thecommunication delays affect interprocess communication, knowledge of the sensitivity ofthe delays to the parameters can be a useful aid in designing a high performance parallelsystem. To keep the analysis tractable, an approximate Markov model is used thatrequires the use of fixed-point iteration for its solution. Validation of the results against asimulation study reveals that the analysis predicts the performance of the network withhigh accuracy.