Journal of Parallel and Distributed Computing
Noncontiguous Processor Allocation Algorithms for Mesh-Connected Multicomputers
IEEE Transactions on Parallel and Distributed Systems
Performance improvement of allocation schemes for mesh-connected computers
Journal of Parallel and Distributed Computing
An Efficient Submesh Allocation Scheme for Two-Dimensional Meshes with Little Overhead
IEEE Transactions on Parallel and Distributed Systems
Introduction to Parallel Computing
Introduction to Parallel Computing
Allocating Precise Submeshes in Mesh Connected Systems
IEEE Transactions on Parallel and Distributed Systems
Job Scheduling is More Important than Processor Allocation for Hypercube Computers
IEEE Transactions on Parallel and Distributed Systems
Adaptive Multicast Wormhole Routing in 2D Mesh Multicomputers
PARLE '93 Proceedings of the 5th International PARLE Conference on Parallel Architectures and Languages Europe
The Effects of Network Contention on Processor Allocation Strategies
IPPS '96 Proceedings of the 10th International Parallel Processing Symposium
ICPADS '06 Proceedings of the 12th International Conference on Parallel and Distributed Systems - Volume 2
An efficient free-list submesh allocation scheme for two-dimensional mesh-connected multicomputers
Journal of Systems and Software
An Adaptive Submesh Allocation Strategy for Two-Dimensional Mesh Connected Systems
ICPP '93 Proceedings of the 1993 International Conference on Parallel Processing - Volume 02
PAAP '10 Proceedings of the 2010 3rd International Symposium on Parallel Architectures, Algorithms and Programming
Hi-index | 0.00 |
Fragmentation can significantly limit the performance of mesh-based multicomputer systems. The adaptive non-contiguous allocation (ANCA) strategy solves the problem of fragmentation by allowing parallel jobs to be allocated non-contiguously. This is done by splitting the frame requested by the parallel job in hand into two subframes of equal sizes at the longest dimension of the request. This may result in having internal fragmentation problem. To remedy this problem, the ANCA strategy uses bookkeeping to keep tracking idle nodes. In this paper, the ANCA strategy is revisited with a new implementation and with more exhaustive simulation-based evaluation. Further, the idea of preventing over-splitting or over-partitioning of parallel requests is studied with the ANCA algorithm. In the proposed implementation, over-partitioning is avoided by placing a limit to maximum number of non-contiguous blocks that can be assigned to any parallel job. This maximum number is referred to as the partitioning-bound. Our experimental results show that having this partitioning-bound parameter can make the ANCA allocation strategy flexible and tunable as it allows the allocator module to choose an optimal partitioning-bound value while allowing parallel jobs to be allocated early without having them over-partitioned. Our experimental results also showed that the ANCA allocation strategy could sustain higher system and communication loads compared to other non-contiguous allocation strategies, namely; the MBS and Paging allocation strategies.