Executing a Program on the MIT Tagged-Token Dataflow Architecture
IEEE Transactions on Computers
Performance Evaluation of a Dataflow Architecture
IEEE Transactions on Computers
Adaptation in natural and artificial systems
Adaptation in natural and artificial systems
Automated learning of load-balancing strategies for a distributed computer system
Automated learning of load-balancing strategies for a distributed computer system
On Runtime Parallel Scheduling for Processor Load Balancing
IEEE Transactions on Parallel and Distributed Systems
Strategies for Dynamic Load Balancing on Highly Parallel Computers
IEEE Transactions on Parallel and Distributed Systems
Practical Dynamic Load Balancing for Irregular Problems
IRREGULAR '96 Proceedings of the Third International Workshop on Parallel Algorithms for Irregularly Structured Problems
Adaptive load balancing: a study in multi-agent learning
Journal of Artificial Intelligence Research
| Hi-index | 0.00 |
A fully scalable, cellular multiprocessor architecture is proposed that is able to dynamically adapt its processing resources to varying demands of signal processing applications. This ability is achieved by migration of tasks between processor cells at run-time such as to avoid cell overload. Several dynamic migration strategies are investigated, and simulation results are provided for different load cases. The results indicate a potential performance gain from dynamic task migration on signal processing applications. By employing a rule-based learning system for an adaptive combination of migration strategies, the migration benefits become independent from the particular application characteristics.